Search

The subtleties of human perception, as measured by vision scientists through the use of psychophysics, are important clues to the internal workings of visual recognition. For instance, measured reaction time can indicate whether a visual stimulus is easy for a subject to recognize, or whether it is hard. In this paper, we consider how to incorporate psychophysical measurements of visual perception into the loss function of a deep neural network being trained for a recognition task, under the assumption that such information can enforce consistency with human behavior. As a case study to assess the viability of this approach, we look at the problem of handwritten document transcription. While good progress has been made towards automatically transcribing modern handwriting, significant challenges remain in transcribing historical documents. Here we describe a general enhancement strategy, underpinned by the new loss formulation, which can be applied to the training regime of any deep learning-based document transcription system. Through experimentation, reliable performance improvement is demonstrated for the standard IAM and RIMES datasets for three different network architectures. Further, we go on to show feasibility for our approach on a new dataset of digitized Latin manuscripts, originally produced by scribes in the Cloister of St. Gall in the the 9th century.

From QAnon conspiracy theories to Russian government sponsored election interference, social media disinformation campaigns are a part of online life, and identifying these threats amid the posts t...

Biometric recognition technology has made significant advances over the last decade and is now used across a number of services and applications. However, this widespread deployment has also resulted in privacy concerns and evolving societal expectations about the appropriate use of the technology. For example, the ability to automatically extract age, gender, race, and health cues from biometric data has heightened concerns about privacy leakage. Face recognition technology, in particular, has been in the spotlight, and is now seen by many as posing a considerable risk to personal privacy. In response to these and similar concerns, researchers have intensified efforts towards developing techniques and computational models capable of ensuring privacy to individuals, while still facilitating the utility of face recognition technology in several application scenarios. These efforts have resulted in a multitude of privacy–enhancing techniques that aim at addressing privacy risks originating from biometric systems and providing technological solutions for legislative requirements set forth in privacy laws and regulations, such as GDPR. The goal of this overview paper is to provide a comprehensive introduction into privacy–related research in the area of biometrics and review existing work on Biometric Privacy–Enhancing Techniques (B–PETs) applied to face biometrics. To make this work useful for as wide of an audience as possible, several key topics are covered as well, including evaluation strategies used with B–PETs, existing datasets, relevant standards, and regulations and critical open issues that will have to be addressed in the future.

In this paper we address the problem of hallucinating high-resolution facial images from low-resolution inputs at high magnification factors. We approach this task with convolutional neural networks (CNNs) and propose a novel (deep) face hallucination model that incorporates identity priors into the learning procedure. The model consists of two main parts: i) a cascaded super-resolution network that upscales the low-resolution facial images, and ii) an ensemble of face recognition models that act as identity priors for the super-resolution network during training. Different from most competing super-resolution techniques that rely on a single model for upscaling (even with large magnification factors), our network uses a cascade of multiple SR models that progressively upscale the low-resolution images using steps of $2\times $ . This characteristic allows us to apply supervision signals (target appearances) at different resolutions and incorporate identity constraints at multiple-scales. The proposed C-SRIP model (Cascaded Super Resolution with Identity Priors) is able to upscale (tiny) low-resolution images captured in unconstrained conditions and produce visually convincing results for diverse low-resolution inputs. We rigorously evaluate the proposed model on the Labeled Faces in the Wild (LFW), Helen and CelebA datasets and report superior performance compared to the existing state-of-the-art.

In a period of fewer than 10 years, the quest for self-driving vehicles, also referred to as autonomous vehicles (AVs) or driverless cars, has become one of the biggest technology races in the world, with tens of billions of dollars poured into companies and start-ups. The goal is an on-road, consumer-driverless car: whether owned by individuals or part of a centralized ride-sharing fleet, this is the area where the majority of investment has occurred. However, AVs have been around for much longer in other fields, such as mining, which share some but not all of the same technical challenges faced by on-road AVs. In this article, we provide an overview of the key technical challenges and solutions for both onand off-road AVs, with a focus on one of the key unsolved challenges-interaction with vulnerable road users (VRUs).

By providing substantial amounts of data and standardized evaluation protocols, datasets in computer vision have helped fuel advances across all areas of visual recognition. But even in light of breakthrough results on recent benchmarks, it is still fair to ask if our recognition algorithms are doing as well as we think they are. The vision sciences at large make use of a very different evaluation regime known as Visual Psychophysics to study visual perception. Psychophysics is the quantitative examination of the relationships between controlled stimuli and the behavioral responses they elicit in experimental test subjects. Instead of using summary statistics to gauge performance, psychophysics directs us to construct item-response curves made up of individual stimulus responses to find perceptual thresholds, thus allowing one to identify the exact point at which a subject can no longer reliably recognize the stimulus class. In this article, we introduce a comprehensive evaluation framework for visual recognition models that is underpinned by this methodology. Over millions of procedurally rendered 3D scenes and 2D images, we compare the performance of well-known convolutional neural networks. Our results bring into question recent claims of human-like performance, and provide a path forward for correcting newly surfaced algorithmic deficiencies., Comment: 9 pages, 4 figures. Published at IEEE Transactions on Pattern Analysis and Machine Intelligence. For supplemental material see http://bjrichardwebster.com/papers/psyphy/supp

Computer vision approaches are widely used by autonomous robotic systems to sense the world around them and to guide their decision making as they perform diverse tasks such as collision avoidance, search and rescue, and object manipulation. High accuracy is critical, particularly for Human-on-the-loop (HoTL) systems where decisions are made autonomously by the system, and humans play only a supervisory role. Failures of the vision model can lead to erroneous decisions with potentially life or death consequences. In this paper, we propose a solution based upon adaptive autonomy levels, whereby the system detects loss of reliability of these models and responds by temporarily lowering its own autonomy levels and increasing engagement of the human in the decision-making process. Our solution is applicable for vision-based tasks in which humans have time to react and provide guidance. When implemented, our approach would estimate the reliability of the vision task by considering uncertainty in its model, and by performing covariate analysis to determine when the current operating environment is ill-matched to the model's training data. We provide examples from DroneResponse, in which small Unmanned Aerial Systems are deployed for Emergency Response missions, and show how the vision model's reliability would be used in addition to confidence scores to drive and specify the behavior and adaptation of the system's autonomy. This workshop paper outlines our proposed approach and describes open challenges at the intersection of Computer Vision and Software Engineering for the safe and reliable deployment of vision models in the decision making of autonomous systems.

The analysis of fish behavior in response to odor stimulation is a crucial component of the general study of cross-modal sensory integration in vertebrates. In zebrafish, the centrifugal pathway runs between the olfactory bulb and the neural retina, originating at the terminalis neuron in the olfactory bulb. Any changes in the ambient odor of a fish’s environment warrants a change in visual sensitivity and can trigger mating-like behavior in males due to increased GnRH signaling in the terminalis neuron. Behavioral experiments to study this phenomenon are commonly conducted in a controlled environment where a video of the fish is recorded over time before and after the application of chemicals to the water. Given the subtleties of behavioral change, trained biologists are currently required to annotate such videos as part of a study. This process of manually analyzing the videos is time-consuming, requires multiple experts to avoid human error/bias and cannot be easily crowdsourced on the Internet. Machine learning algorithms from computer vision, on the other hand, have proven to be effective for video annotation tasks because they are fast, accurate, and, if designed properly, can be less biased than humans. In this work, we propose to automate the entire process of analyzing videos of behavior changes in zebrafish by using tools from computer vision, relying on minimal expert supervision. The overall objective of this work is to create a generalized tool to predict animal behaviors from videos using state-of-the-art deep learning models, with the dual goal of advancing understanding in biology and engineering a more robust and powerful artificial information processing system for biologists.

Understanding the structure of complex activities in untrimmed videos is a challenging task in the area of action recognition. One problem here is that this task usually requires a large amount of hand-annotated minute- or even hour-long video data, but annotating such data is very time consuming and can not easily be automated or scaled. To address this problem, this paper proposes an approach for the unsupervised learning of actions in untrimmed video sequences based on a joint visual-temporal embedding space. To this end, we combine a visual embedding based on a predictive U-Net architecture with a temporal continuous function. The resulting representation space allows detecting relevant action clusters based on their visual as well as their temporal appearance. The proposed method is evaluated on three standard benchmark datasets, Breakfast Actions, INRIA YouTube Instructional Videos, and 50 Salads. We show that the proposed approach is able to provide a meaningful visual and temporal embedding out of the visual cues present in contiguous video frames and is suitable for the task of unsupervised temporal segmentation of actions.

This paper introduces an agent-centric approach to handle novelty in the visual recognition domain of handwriting recognition (HWR). An ideal transcription agent would rival or surpass human perception, being able to recognize known and new characters in an image, and detect any stylistic changes that may occur within or across documents. A key confound is the presence of novelty, which has continued to stymie even the best machine learning-based algorithms for these tasks. In handwritten documents, novelty can be a change in writer, character attributes, writing attributes, or overall document appearance, among other things. Instead of looking at each aspect independently, we suggest that an integrated agent that can process known characters and novelties simultaneously is a better strategy. This paper formalizes the domain of handwriting recognition with novelty, describes a baseline agent, introduces an evaluation protocol with benchmark data, and provides experimentation to set the state-of-the-art. Results show feasibility for the agent-centric approach, but more work is needed to approach human-levels of reading ability, giving the HWR community a formal basis to build upon as they solve this challenging problem.

Computer vision has achieved superior results with the rapid development of new techniques in deep neural networks. Object detection in the wild is a core task in computer vision, and already has many successful applications in the real world. However, deep neural networks for object detection usually consist of hundreds, and sometimes even thousands, of layers. Training such networks is challenging, and training data has a fundamental impact on model performance. Because data collection and annotation are expensive and labor-intensive, lots of data augmentation methods have been proposed to generate synthetic data for neural network training. Most of those methods focus on manipulating 2D images. In contrast to that, in this paper, we leverage the realistic visual effects of 3D environments and propose a new way of generating synthetic data for computer vision tasks related to city scenes. Specifically, we describe a pipeline that can generate a 3D city model from an input of a 2D image that portrays the layout design of a city. This pipeline also takes optional parameters to further customize the output 3D city model. Using our pipeline, a virtual 3D city model with high-quality textures can be generated within seconds, and the output is an object ready to render. The model generated will assist people with limited 3D development knowledge to create high quality city scenes for different needs. As examples, we show the use of generated 3D city models as the synthetic data source for a scene text detection task and a traffic sign detection task. Both qualitative and quantitative results show that the generated virtual city is a good match to real-world data and potentially can benefit other computer vision tasks with similar contexts.

Computer Vision practitioners must thoroughly understand their model's performance, but conditional evaluation is complex and error-prone. In biometric verification, model performance over continuous covariates - known, real-number attributes of images that affect performance - is particularly challenging to study. We develop a generative model of the match and non-match score distributions over continuous covariates and perform inference with modern Bayesian methods. We use mixture models to capture arbitrary distributions and local basis functions to capture non-linear, multivariate trends. Three experiments demonstrate the accuracy and effectiveness of our approach. First, we study the relationship between age and face verification performance and find previous methods may overstate performance and confidence. Second, we study preprocessing for CNNs and find a highly non-linear, multivariate surface of model performance. Our method is accurate and data efficient when evaluated against previous synthetic methods. Third, we demonstrate the novel application of our method to pedestrian tracking and calculate variable thresholds and expected performance while controlling for multiple covariates.

We present a new white-box backdoor attack that exploits a vulnerability of convolutional neural networks (CNNs). In particular, we examine the application of facial recognition. Deep learning techniques are at the top of the game for facial recognition, which means they have now been implemented in many production-level systems. Alarmingly, unlike other commercial technologies such as operating systems and network devices, deep learning-based facial recognition algorithms are not presently designed with security requirements or audited for security vulnerabilities before deployment. Given how young the technology is and how abstract many of the internal workings of these algorithms are, neural network-based facial recognition systems are prime targets for security breaches. As more and more of our personal information begins to be guarded by facial recognition (e.g., the iPhone X), exploring the security vulnerabilities of these systems from a penetration testing standpoint is crucial. Along these lines, we describe a general methodology for backdooring CNNs via targeted weight perturbations. Using a five-layer CNN and ResNet-50 as case studies, we show that an attacker is able to significantly increase the chance that inputs they supply will be falsely accepted by a CNN while simultaneously preserving the error rates for legitimate enrolled classes.

An interesting development in automatic visual recognition has been the emergence of tasks where it is not possible to assign objective labels to images, yet still feasible to collect annotations that reflect human judgements about them. Machine learning-based predictors for these tasks rely on supervised training that models the behavior of the annotators, i.e., what would the average person's judgement be for an image? A key open question for this type of work, especially for applications where inconsistency with human behavior can lead to ethical lapses, is how to evaluate the epistemic uncertainty of trained predictors, i.e., the uncertainty that comes from the predictor's model. We propose a Bayesian framework for evaluating black box predictors in this regime, agnostic to the predictor's internal structure. The framework specifies how to estimate the epistemic uncertainty that comes from the predictor with respect to human labels by approximating a conditional distribution and producing a credible interval for the predictions and their measures of performance. The framework is successfully applied to four image classification tasks that use subjective human judgements: facial beauty assessment, social attribute assignment, apparent age estimation, and ambiguous scene labeling., 21 pages. 6 figures. 2 tables. Supplementary Material as Appendix with 28 pages, 6 figures, 2 tables. First major revision for journal Pattern Recognition. Code to be included after publication at https://github.com/prijatelj/bayesian_eval_ground_truth-free

This article contains examples of offensive and abusive language as a necessary part of illustrating its research findings. The automatic analysis of face images can generate predictions about a person’s gender, age, race, facial expression, body mass index, and various other indices and conditions. A few recent publications have claimed success in analyzing an image of a person’s face in order to predict the person’s status as Criminal/Noncriminal. Predicting “criminality from face” may initially seem similar to other facial analytics, but we argue that attempts to create a criminality-from-face algorithm are necessarily doomed to fail, that apparently promising experimental results in recent publications are an illusion resulting from inadequate experimental design, and that there is potentially a large social cost to belief in the criminality from face illusion.

Imaging is a dominant strategy for data collection in neuroscience, yielding stacks of images that often scale to gigabytes of data for a single experiment. Machine learning algorithms from computer vision can serve as a pair of virtual eyes that tirelessly processes these images, automatically detecting and identifying microstructures. Unlike learning methods, our Flexible Learning-free Reconstruction of Imaged Neural volumes (FLoRIN) pipeline exploits structure-specific contextual clues and requires no training. This approach generalizes across different modalities, including serially-sectioned scanning electron microscopy (sSEM) of genetically labeled and contrast enhanced processes, spectral confocal reflectance (SCoRe) microscopy, and high-energy synchrotron X-ray microtomography (μCT) of large tissue volumes. We deploy the FLoRIN pipeline on newly published and novel mouse datasets, demonstrating the high biological fidelity of the pipeline’s reconstructions. FLoRIN reconstructions are of sufficient quality for preliminary biological study, for example examining the distribution and morphology of cells or extracting single axons from functional data. Compared to existing supervised learning methods, FLoRIN is one to two orders of magnitude faster and produces high-quality reconstructions that are tolerant to noise and artifacts, as is shown qualitatively and quantitatively.

The application of deep learning in robotics leads to very specific problems and research questions that are typically not addressed by the computer vision and machine learning communities. In this paper we discuss a number of robotics-specific learning, reasoning, and embodiment challenges for deep learning. We explain the need for better evaluation metrics, highlight the importance and unique challenges for deep robotic learning in simulation, and explore the spectrum between purely data-driven and model-driven approaches. We hope this paper provides a motivating overview of important research directions to overcome the current limitations, and helps to fulfill the promising potentials of deep learning in robotics.

The ease with which one can edit and redistribute digital documents on the Internet is one of modernity's great achievements, but it also leads to some vexing problems. With growing academic interest in the study of the evolution of digital writing on the one hand and the rise of disinformation on the other, the problem of identifying the relationship between texts with similar content is becoming more important. Traditional vector space representations of texts have made progress in solving this problem when it is cast as a reconstruction task that organizes related texts into a tree expressing relationships-this is dubbed text phylogeny in the information forensics literature. However, as new text representation methods have been successfully applied to many other text analysis problems, it is worth investigating if they too are used in text phylogeny tree reconstruction. In this paper, we explore the use of word embeddings as a text representation method, with the aim of trying to improve the accuracy of reconstructed phylogeny trees for real-world data and compare it with other widely used text representation methods. We evaluate the performance on established benchmarks for this task: a synthetic data set and data collected from Wikipedia. We also apply our framework to a new data set of fan fiction based on some famous fairy tales. Experimental results show that word embeddings are competitive with other feature sets for the published benchmarks, and are highly effective for creative writing.

The veil of anonymity provided by smartphones with pre-paid SIM cards, public Wi-Fi hotspots, and distributed networks like Tor has drastically complicated the task of identifying users of social media during forensic investigations. In some cases, the text of a single posted message will be the only clue to an author’s identity. How can we accurately predict who that author might be when the message may never exceed 140 characters on a service like Twitter? For the past 50 years, linguists, computer scientists, and scholars of the humanities have been jointly developing automated methods to identify authors based on the style of their writing. All authors possess peculiarities of habit that influence the form and content of their written works. These characteristics can often be quantified and measured using machine learning algorithms. In this paper, we provide a comprehensive review of the methods of authorship attribution that can be applied to the problem of social media forensics. Furthermore, we examine emerging supervised learning-based methods that are effective for small sample sizes, and provide step-by-step explanations for several scalable approaches as instructional case studies for newcomers to the field. We argue that there is a significant need in forensics for new authorship attribution algorithms that can exploit context, can process multi-modal data, and are tolerant to incomplete knowledge of the space of all possible authors at training time.

The most straightforward, yet insecure, methods of authenticating a person on smartphones derive from the solutions applied to personal computers or smart cards, namely the authorization by passwords or numeric codes. Alarmingly, the widespread use of smartphone platforms implies that people are carrying around sensitive information in their pocket, making the information more available physically. As smartphone owners are often using their devices in public areas, these short numeric codes or other forms of passwords can be obtained quickly through shoulder surfing, resulting in making that restricted data far more accessible for those who are not authorized to access the device. In this paper, we address the problem of biometric verifi-cation on smartphones. We propose a new approach for gesture-based verification that makes use of open set recognition algorithms. Further, we introduce a new database of inertial measurements to investigate the user identification capabilities of this approach. The results we have obtained indicate that this approach is a feasible solution, although the precision of the method depends highly on the chosen samples of the training set.

What is the current state-of-the-art for image restoration and enhancement applied to degraded images acquired under less than ideal circumstances? Can the application of such algorithms as a pre-processing step to improve image interpretability for manual analysis or automatic visual recognition to classify scene content? While there have been important advances in the area of computational photography to restore or enhance the visual quality of an image, the capabilities of such techniques have not always translated in a useful way to visual recognition tasks. Consequently, there is a pressing need for the development of algorithms that are designed for the joint problem of improving visual appearance and recognition, which will be an enabling factor for the deployment of visual recognition tools in many real-world scenarios. To address this, we introduce the UG^2 dataset as a large-scale benchmark composed of video imagery captured under challenging conditions, and two enhancement tasks designed to test algorithmic impact on visual quality and automatic object recognition. Furthermore, we propose a set of metrics to evaluate the joint improvement of such tasks as well as individual algorithmic advances, including a novel psychophysics-based evaluation regime for human assessment and a realistic set of quantitative measures for object recognition performance. We introduce six new algorithms for image restoration or enhancement, which were created as part of the IARPA sponsored UG^2 Challenge workshop held at CVPR 2018. Under the proposed evaluation regime, we present an in-depth analysis of these algorithms and a host of deep learning-based and classic baseline approaches. From the observed results, it is evident that we are in the early days of building a bridge between computational photography and visual recognition, leaving many opportunities for innovation in this area., Comment: CVPR Prize Challenge: http://www.ug2challenge.org

We present and share a foundational dataset of multi-angle video recordings of scripted athletic movements to enable the development of computer vision research applications that evaluate and identify lower-body injury risk. The focus of the dataset is female athletes, who are at a substantially increased risk of anterior cruciate ligament (ACL) injury and are therefore a top priority for sports science. In our study, varsity and club sport athletes perform two assessment movements (the countermovement jump and the drop jump). These jump tasks are used ubiquitously in sports medicine research to characterize athleticism and to identify risk factors that indicate ACL injury propensity. The novelty of the dataset centers on (i) the type of movement data (purposeful, evaluative movements that need to be tracked with a high degree of precision), (ii) our generalized collection method that can be replicated with ease by non-experts, and (iii) the amount of data collected (we collected data from 55 division one (D1) female athletes performing 3-5 iterations of each jumps, for a total of 480 jumps). Data from each camera was manually aligned and a fully automated pipeline was built to extract knee information from athletes. Ideally, any athlete or researcher will be able to easily replicate our setup and assemble a compatible and complementary dataset to propel the development and assessment of injury propensity models.

Neural sequence-to-sequence models, particularly the Transformer, are the state of the art in machine translation. Yet these neural networks are very sensitive to architecture and hyperparameter settings. Optimizing these settings by grid or random search is computationally expensive because it requires many training runs. In this paper, we incorporate architecture search into a single training run through auto-sizing, which uses regularization to delete neurons in a network over the course of training. On very low-resource language pairs, we show that auto-sizing can improve BLEU scores by up to 3.9 points while removing one-third of the parameters from the model., The 3rd Workshop on Neural Generation and Translation (WNGT 2019)

Subject matching performance in iris biometrics is contingent upon fast, high-quality iris segmentation. In many cases, iris biometrics acquisition equipment takes a number of images in sequence and combines the segmentation and matching results for each image to strengthen the result. To date, segmentation has occurred in 2D, operating on each image individually. But such methodologies, while powerful, do not take advantage of potential gains in performance afforded by treating sequential images as volumetric data. As a first step in this direction, we apply the Flexible Learning-Free Reconstructoin of Neural Volumes (FLoRIN) framework, an open source segmentation and reconstruction framework originally designed for neural microscopy volumes, to volumetric segmentation of iris videos. Further, we introduce a novel dataset of near-infrared iris videos, in which each subject's pupil rapidly changes size due to visible-light stimuli, as a test bed for FLoRIN. We compare the matching performance for iris masks generated by FLoRIN, deep-learning-based (SegNet), and Daugman's (OSIRIS) iris segmentation approaches. We show that by incorporating volumetric information, FLoRIN achieves a factor of 3.6 to an order of magnitude increase in throughput with only a minor drop in subject matching performance. We also demonstrate that FLoRIN-based iris segmentation maintains this speedup on low-resource hardware, making it suitable for embedded biometrics systems., Comment: 9 pages, 4 figures

In many instances, one piece of evidence alone will not be enough to make an accurate decision about the intertextual relationship between two texts. More evidence, combined in a probabilistic way, can improve our confidence in a classification result.

Creative works, whether paintings or memes, follow unique journeys that result in their final form. Understanding these journeys, a process known as "provenance analysis", provides rich insights into the use, motivation, and authenticity underlying any given work. The application of this type of study to the expanse of unregulated content on the Internet is what we consider in this paper. Provenance analysis provides a snapshot of the chronology and validity of content as it is uploaded, re-uploaded, and modified over time. Although still in its infancy, automated provenance analysis for online multimedia is already being applied to different types of content. Most current works seek to build provenance graphs based on the shared content between images or videos. This can be a computationally expensive task, especially when considering the vast influx of content that the Internet sees every day. Utilizing non-content-based information, such as timestamps, geotags, and camera IDs can help provide important insights into the path a particular image or video has traveled during its time on the Internet without large computational overhead. This paper tests the scope and applicability of metadata-based inferences for provenance graph construction in two different scenarios: digital image forensics and cultural analytics., Supplemental material for this paper can be found at https://drive.google.com/file/d/1Tbs2CQg_VQAc2PdztW5twVaiXD0G12-H/view?usp=sharing

The diverse phenomena which have been treated under the label intertextuality can all be seen as aspects of a general human ability to extract statistical patterns from complex data.

Contemporary face hallucination (FH) models exhibit considerable ability to reconstruct high-resolution (HR) details from low-resolution (LR) face images. This ability is commonly learned from examples of corresponding HR-LR image pairs, created by artificially down-sampling the HR ground truth data. This down-sampling (or degradation) procedure not only defines the characteristics of the LR training data, but also determines the type of image degradations the learned FH models are eventually able to handle. If the image characteristics encountered with real-world LR images differ from the ones seen during training, FH models are still expected to perform well, but in practice may not produce the desired results. In this paper we study this problem and explore the bias introduced into FH models by the characteristics of the training data. We systematically analyze the generalization capabilities of several FH models in various scenarios, where the image the degradation function does not match the training setup and conduct experiments with synthetically downgraded as well as real-life low-quality images. We make several interesting findings that provide insight into existing problems with FH models and point to future research directions.

We propose an algorithm to generate realistic face images of both real and synthetic identities (people who do not exist) with different facial yaw, shape and resolution.The synthesized images can be used to augment datasets to train CNNs or as massive distractor sets for biometric verification experiments without any privacy concerns. Additionally, law enforcement can make use of this technique to train forensic experts to recognize faces. Our method samples face components from a pool of multiple face images of real identities to generate the synthetic texture. Then, a real 3D head model compatible to the generated texture is used to render it under different facial yaw transformations. We perform multiple quantitative experiments to assess the effectiveness of our synthesis procedure in CNN training and its potential use to generate distractor face images. Additionally, we compare our method with popular GAN models in terms of visual quality and execution time., To appear in IEEE WACV 2019. Get our data (2M face images of 12K synthetic subjects, 8K 3D head models) by accessing the "Notre Dame Synthetic Face Dataset" here: https://cvrl.nd.edu/projects/data/

Imaging is a dominant strategy for data collection in neuroscience, yielding stacks of images that often scale to gigabytes of data for a single experiment. Machine learning algorithms from computer vision can serve as a pair of virtual eyes that tirelessly processes these images, automatically detecting and identifying microstructures. Unlike learning methods, our Flexible Learning-free Reconstruction of Imaged Neural volumes (FLoRIN) pipeline exploits structure-specific contextual clues and requires no training. This approach generalizes across different modalities, including serially-sectioned scanning electron microscopy (sSEM) of genetically labeled and contrast enhanced processes, spectral confocal reflectance (SCoRe) microscopy, and high-energy synchrotron X-ray microtomography (μCT) of large tissue volumes. We deploy the FLoRIN pipeline on newly published and novel mouse datasets, demonstrating the high biological fidelity of the pipeline's reconstructions. FLoRIN reconstructions are of sufficient quality for preliminary biological study, for example examining the distribution and morphology of cells or extracting single axons from functional data. Compared to existing supervised learning methods, FLoRIN is one to two orders of magnitude faster and produces high-quality reconstructions that are tolerant to noise and artifacts, as is shown qualitatively and quantitatively.

Neuroscience theory posits that the brain's visual system coarsely identifies broad object categories via neural activation patterns, with similar objects producing similar neural responses. Artificial neural networks also have internal activation behavior in response to stimuli. We hypothesize that networks exhibiting brain-like activation behavior will demonstrate brain-like characteristics, e.g., stronger generalization capabilities. In this paper we introduce a human-model similarity (HMS) metric, which quantifies the similarity of human fMRI and network activation behavior. To calculate HMS, representational dissimilarity matrices (RDMs) are created as abstractions of activation behavior, measured by the correlations of activations to stimulus pairs. HMS is then the correlation between the fMRI RDM and the neural network RDM across all stimulus pairs. We test the metric on unsupervised predictive coding networks, which specifically model visual perception, and assess the metric for statistical significance over a large range of hyperparameters. Our experiments show that networks with increased human-model similarity are correlated with better performance on two computer vision tasks: next frame prediction and object matching accuracy. Further, HMS identifies networks with high performance on both tasks. An unexpected secondary finding is that the metric can be employed during training as an early-stopping mechanism., Under review

Machine learning is a field of computer science that builds algorithms that learn. In many cases, machine learning algorithms are used to recreate a human ability like adding a caption to a photo, driving a car, or playing a game. While the human brain has long served as a source of inspiration for machine learning, little effort has been made to directly use data collected from working brains as a guide for machine learning algorithms. Here we demonstrate a new paradigm of “neurally-weighted” machine learning, which takes fMRI measurements of human brain activity from subjects viewing images, and infuses these data into the training process of an object recognition learning algorithm to make it more consistent with the human brain. After training, these neurally-weighted classifiers are able to classify images without requiring any additional neural data. We show that our neural-weighting approach can lead to large performance gains when used with traditional machine vision features, as well as to significant improvements with already high-performing convolutional neural network features. The effectiveness of this approach points to a path forward for a new class of hybrid machine learning algorithms which take both inspiration and direct constraints from neuronal data.

Imaging is a dominant strategy for data collection in neuroscience, yielding stacks of images that often scale to gigabytes of data for a single experiment. Machine learning algorithms from computer vision can serve as a pair of virtual eyes that tirelessly processes these images, automatically constructing more complete and realistic circuits. In practice, such algorithms are often too error-prone and computationally expensive to be immediately useful. We address these shortcomings with a new fast, flexible, learning-free method for sparse segmentation and reconstruction of neural volumes. Unlike learning methods, our Flexible Learning-free Reconstruction of Imaged Neural volumes (FLoRIN) pipeline exploits structure-specific contextual clues and requires no training. This approach generalizes across different modalities, including serially-sectioned scanning electron microscopy (sSEM) of genetically labeled and contrast enhanced processes, spectral confocal reflectance (SCoRe) microscopy, and high-energy synchrotron X-ray microtomography (μCT) of large tissue volumes. We deploy the FLoRIN pipeline on newly published and novel mouse datasets, demonstrating the high biological fidelity of the pipeline’s reconstructions, which are of sufficient quality for preliminary biological study. Compared to existing supervised learning methods, it is both significantly faster (up to several orders of magnitude) and produces high-quality reconstructions that are robust to noise and artifacts.

Online writers and journalism media are increasingly combining visualization (and other multimedia content) with narrative text to create narrative visualizations. Often, however, the two elements are presented independently of one another. We propose an approach to automatically integrate text and visualization elements. We begin with a writer's narrative that presumably can be supported with visual data evidence. We leverage natural language processing, quantitative narrative analysis, and information visualization to (1) automatically extract narrative components (who, what, when, where) from data-rich stories, and (2) integrate the supporting data evidence with the text to develop a narrative visualization. We also employ bidirectional interaction from text to visualization and visualization to text to support reader exploration in both directions. We demonstrate the approach with a case study in the data-rich field of sports journalism., Comment: ACM IUI'18, 3 figures, 5 pages

Face recognition performance has improved remarkably in the last decade. Much of this success can be attributed to the development of deep learning techniques such as convolutional neural networks (CNNs). While CNNs have pushed the state-of-the-art forward, their training process requires a large amount of clean and correctly labelled training data. If a CNN is intended to tolerate facial pose, then we face an important question: should this training data be diverse in its pose distribution, or should face images be normalized to a single pose in a pre-processing step? To address this question, we evaluate a number of popular facial landmarking and pose correction algorithms to understand their effect on facial recognition performance. Additionally, we introduce a new, automatic, single-image frontalization scheme that exceeds the performance of current algorithms. CNNs trained using sets of different pre-processing methods are used to extract features from the Point and Shoot Challenge (PaSC) and CMU Multi-PIE datasets. We assert that the subsequent verification and recognition performance serves to quantify the effectiveness of each pose correction scheme., Comment: Accepted to WACV 2018 - Fixed title to correct working version Code available here: https://github.com/joelb92/ND_Frontalization_Project/tree/master/Release

We propose a multi-scale GAN model to hallucinate realistic context (forehead, hair, neck, clothes) and background pixels automatically from a single input face mask. Instead of swapping a face on to an existing picture, our model directly generates realistic context and background pixels based on the features of the provided face mask. Unlike face inpainting algorithms, it can generate realistic hallucinations even for a large number of missing pixels. Our model is composed of a cascaded network of GAN blocks, each tasked with hallucination of missing pixels at a particular resolution while guiding the synthesis process of the next GAN block. The hallucinated full face image is made photo-realistic by using a combination of reconstruction, perceptual, adversarial and identity preserving losses at each block of the network. With a set of extensive experiments, we demonstrate the effectiveness of our model in hallucinating context and background pixels from face masks varying in facial pose, expression and lighting, collected from multiple datasets subject disjoint with our training data. We also compare our method with two popular face swapping and face completion methods in terms of visual quality and recognition performance. Additionally, we analyze our cascaded pipeline and compare it with the recently proposed progressive growing of GANs., Comment: Extended version of WACV 2020 paper

In the last decade, video blogs (vlogs) have become an extremely popular method through which people express sentiment. The ubiquitousness of these videos has increased the importance of multimodal fusion models, which incorporate video and audio features with traditional text features for automatic sentiment detection. Multimodal fusion offers a unique opportunity to build models that learn from the full depth of expression available to human viewers. In the detection of sentiment in these videos, acoustic and video features provide clarity to otherwise ambiguous transcripts. In this paper, we present a multimodal fusion model that exclusively uses high-level video and audio features to analyze spoken sentences for sentiment. We discard traditional transcription features in order to minimize human intervention and to maximize the deployability of our model on at-scale real-world data. We select high-level features for our model that have been successful in nonaffect domains in order to test their generalizability in the sentiment detection domain. We train and test our model on the newly released CMU Multimodal Opinion Sentiment and Emotion Intensity (CMUMOSEI) dataset, obtaining an F1 score of 0.8049 on the validation set and an F1 score of 0.6325 on the held-out challenge test set., Comment: Published in the First Workshop on Computational Modeling of Human Multimodal Language - ACL 2018

Scientific fields that are interested in faces have developed their own sets of concepts and procedures for understanding how a target model system (be it a person or algorithm) perceives a face under varying conditions. In computer vision, this has largely been in the form of dataset evaluation for recognition tasks where summary statistics are used to measure progress. While aggregate performance has continued to improve, understanding individual causes of failure has been difficult, as it is not always clear why a particular face fails to be recognized, or why an impostor is recognized by an algorithm. Importantly, other fields studying vision have addressed this via the use of visual psychophysics: the controlled manipulation of stimuli and careful study of the responses they evoke in a model system. In this paper, we suggest that visual psychophysics is a viable methodology for making face recognition algorithms more explainable. A comprehensive set of procedures is developed for assessing face recognition algorithm behavior, which is then deployed over state-of-the-art convolutional neural networks and more basic, yet still widely used, shallow and handcrafted feature-based approaches.

In this paper, we propose a novel face synthesis approach that can generate an arbitrarily large number of synthetic images of both real and synthetic identities. Thus a face image dataset can be expanded in terms of the number of identities represented and the number of images per identity using this approach, without the identity-labeling and privacy complications that come from downloading images from the web. To measure the visual fidelity and uniqueness of the synthetic face images and identities, we conducted face matching experiments with both human participants and a CNN pre-trained on a dataset of 2.6M real face images. To evaluate the stability of these synthetic faces, we trained a CNN model with an augmented dataset containing close to 200,000 synthetic faces. We used a snapshot of this trained CNN to recognize extremely challenging frontal (real) face images. Experiments showed training with the augmented faces boosted the face recognition performance of the CNN.

Discriminative learning of Restricted Boltzmann Machines has been recently introduced as an alternative to provide a self-contained approach for both unsupervised feature learning and classification purposes. However, one of the main problems faced by researchers interested in such approach concerns with a proper selection of its parameters, which play an important role in its final performance. In this paper, we introduced some meta-heuristic techniques for this purpose, as well as we showed they can be more accurate than a random search, which is commonly used technique in several works.

Real-world tasks in computer vision often touch upon open set recognition: multi-class recognition with incomplete knowledge of the world and many unknown inputs. Recent work on this problem has proposed a model incorporating an open space risk term to account for the space beyond the reasonable support of known classes. This paper extends the general idea of open space risk limiting classification to accommodate non-linear classifiers in a multiclass setting. We introduce a new open set recognition model called compact abating probability (CAP), where the probability of class membership decreases in value (abates) as points move from known data toward open space. We show that CAP models improve open set recognition for multiple algorithms. Leveraging the CAP formulation, we go on to describe the novel Weibull-calibrated SVM (W-SVM) algorithm, which combines the useful properties of statistical extreme value theory for score calibration with one-class and binary support vector machines. Our experiments show that the W-SVM is significantly better for open set object detection and OCR problems when compared to the state-of-the-art for the same tasks.

In literary study, intertextuality refers to the reuse of text, where new meaning or novel stylistic effects have been generated. Most typically in the digital humanities, algorithms for intertextual analysis search for approximate lexical correspondence that can be described as paraphrase. In this article, we look at a complimentary approach that more closely captures the behavior of the reader when faced with meaningful connections between texts in the absence of words that have the same form or stem, which constrains the match to semantics. The technique we employ for identifying such semantic intertextuality is the popular natural language processing strategy of semantic analysis. Unlike the typical scenario for semantic analysis, where a corpus of long form documents is available, we examine the far more limited textual fragments that embody intertextuality. We are primarily concerned with texts from antiquity, where small phrases or passages often form the locus of comparison. In this vein, we look at a specific case study of established parallels between book 1 of Lucan’s Civil War and all of Vergil’s Aeneid. Applying semantic analysis over these texts, we are able to recover parallels that lexical matching cannot, as well as discover new and interesting thematic matches between the two works.

Learning and recognition is a fundamental process performed in many robot operations such as mapping and localization. The majority of approaches share some common characteristics, such as attempting to extract salient features, landmarks or signatures, and growth in data storage and computational requirements as the size of the environment increases. In biological systems, spatial encoding in the brain is definitively known to be performed using a fixed-size neural encoding framework - the place, head-direction and grid cells found in the mammalian hippocampus and entorhinal cortex. Particularly paradoxically, one of the main encoding centers - the grid cells - represents the world using a highly aliased, repetitive encoding structure where one neuron represents an unbounded number of places in the world. Inspired by this system, in this paper we invert the normal approach used in forming mapping and localization algorithms, by developing a novel place recognition algorithm that seeks out and leverages repetitive, mutually complementary landmark frequencies in the world. The combinatorial encoding capacity of multiple different frequencies enables not only the ability to achieve efficient data storage, but also the potential for sub-linear storage growth in a learning and recall system. Using both ground-based and aerial camera datasets, we demonstrate the system finding and utilizing these frequencies to achieve successful place recognition, and discuss how this approach might scale to arbitrarily large global datasets and dimensions.

In this paper, we consider the problem of automatically segmenting neuronal cells in dual-color confocal microscopy images. This problem is a key task in various quantitative analysis applications in neuroscience, such as tracing cell genesis in Danio rerio (zebrafish) brains. Deep learning, especially using fully convolutional networks (FCN), has profoundly changed segmentation research in biomedical imaging. We face two major challenges in this problem. First, neuronal cells may form dense clusters, making it difficult to correctly identify all individual cells (even to human experts). Consequently, segmentation results of the known FCN-type models are not accurate enough. Second, pixel-wise ground truth is difficult to obtain. Only a limited amount of approximate instance-wise annotation can be collected, which makes the training of FCN models quite cumbersome. We propose a new FCN-type deep learning model, called deep complete bipartite networks (CB-Net), and a new scheme for leveraging approximate instance-wise annotation to train our pixel-wise prediction model. Evaluated using seven real datasets, our proposed new CB-Net model outperforms the state-of-the-art FCN models and produces neuron segmentation results of remarkable quality.

Deriving relationships between images and tracing back their history of modifications are at the core of Multimedia Phylogeny solutions, which aim to combat misinformation through doctored visual media. Nonetheless, most recent image phylogeny solutions cannot properly address cases of forged composite images with multiple donors, an area known as multiple parenting phylogeny (MPP). This paper presents a preliminary undirected graph construction solution for MPP, without any strict assumptions. The algorithm is underpinned by robust image representative keypoints and different geometric consistency checks among matching regions in both images to provide regions of interest for direct comparison. The paper introduces a novel technique to geometrically filter the most promising matches as well as to aid in the shared region localization task. The strength of the approach is corroborated by experiments with real-world cases, with and without image distractors (unrelated cases)., Comment: 5 pages, Accepted in International Conference on Image Processing, 2017

As image tampering becomes ever more sophisticated and commonplace, the need for image forensics algorithms that can accurately and quickly detect forgeries grows. In this paper, we revisit the ideas of image querying and retrieval to provide clues to better localize forgeries. We propose a method to perform large-scale image forensics on the order of one million images using the help of an image search algorithm and database to gather contextual clues as to where tampering may have taken place. In this vein, we introduce five new strongly invariant image comparison methods and test their effectiveness under heavy noise, rotation, and color space changes. Lastly, we show the effectiveness of these methods compared to passive image forensics using Nimble [https://www.nist.gov/itl/iad/mig/nimble-challenge], a new, state-of-the-art dataset from the National Institute of Standards and Technology (NIST)., Comment: 5 pages, 5 figures

Advances in image restoration and enhancement techniques have led to discussion about how such algorithmscan be applied as a pre-processing step to improve automatic visual recognition. In principle, techniques like deblurring and super-resolution should yield improvements by de-emphasizing noise and increasing signal in an input image. But the historically divergent goals of the computational photography and visual recognition communities have created a significant need for more work in this direction. To facilitate new research, we introduce a new benchmark dataset called UG^2, which contains three difficult real-world scenarios: uncontrolled videos taken by UAVs and manned gliders, as well as controlled videos taken on the ground. Over 160,000 annotated frames forhundreds of ImageNet classes are available, which are used for baseline experiments that assess the impact of known and unknown image artifacts and other conditions on common deep learning-based object classification approaches. Further, current image restoration and enhancement techniques are evaluated by determining whether or not theyimprove baseline classification performance. Results showthat there is plenty of room for algorithmic innovation, making this dataset a useful tool going forward., Comment: Supplemental material: https://goo.gl/vVM1xe, Dataset: https://goo.gl/AjA6En, CVPR 2018 Prize Challenge: ug2challenge.org

For robots operating in outdoor environments, a number of factors, including weather, time of day, rough terrain, high speeds, and hardware limitations, make performing vision-based simultaneous localization and mapping with current techniques infeasible due to factors such as image blur and/or underexposure, especially on smaller platforms and low-cost hardware. In this paper, we present novel visual place-recognition and odometry techniques that address the challenges posed by low lighting, perceptual change, and low-cost cameras. Our primary contribution is a novel two-step algorithm that combines fast low-resolution whole image matching with a higher-resolution patch-verification step, as well as image saliency methods that simultaneously improve performance and decrease computing time. The algorithms are demonstrated using consumer cameras mounted on a small vehicle in a mixed urban and vegetated environment and a car traversing highway and suburban streets, at different times of day and night and in various weather conditions. The algorithms achieve reliable mapping over the course of a day, both when incrementally incorporating new visual scenes from different times of day into an existing map, and when using a static map comprising visual scenes captured at only one point in time. Using the two-step place-recognition process, we demonstrate for the first time single-image, error-free place recognition at recall rates above 50% across a day-night dataset without prior training or utilization of image sequences. This place-recognition performance enables topologically correct mapping across day-night cycles.

Recognition is the fundamental task of visual cognition, yet how to formalize the general recognition problem for computer vision remains an open issue. The problem is sometimes reduced to the simplest case of recognizing matching pairs, often structured to allow for metric constraints. However, visual recognition is broader than just pair-matching: what we learn and how we learn it has important implications for effective algorithms. In this review paper, we reconsider the assumption of recognition as a pair-matching test, and introduce a new formal definition that captures the broader context of the problem. Through a meta-analysis and an experimental assessment of the top algorithms on popular data sets, we gain a sense of how often metric properties are violated by recognition algorithms. By studying these violations, useful insights come to light: we make the case for local distances and systems that leverage outside information to solve the general recognition problem.

Camera attribution approaches in digital image forensics have most often been evaluated in a closed set context, whereby all devices are known during training and testing time. However, in a real investigation, we must assume that innocuous images from unknown devices will be recovered, which we would like to remove from the pool of evidence. In pattern recognition, this corresponds to what is known as the open set recognition problem. This article introduces new algorithms for open set modes of image source attribution (identifying whether or not an image was captured by a specific digital camera) and device linking (identifying whether or not a pair of images was acquired from the same digital camera without the need for physical access to the device). Both algorithms rely on a new multi-region feature generation strategy, which serves as a projection space for the class of interest and emphasizes its properties, and on decision boundary carving, a novel method that models the decision space of a trained SVM classifier by taking advantage of a few known cameras to adjust the decision boundaries to decrease false matches from unknown classes. Experiments including thousands of unconstrained images collected from the web show a significant advantage for our approaches over the most competitive prior work.