Search

Compositional Reasoning (CR) entails grasping the significance of attributes, relations, and word order. Recent Vision-Language Models (VLMs), comprising a visual encoder and a Large Language Model (LLM) decoder, have demonstrated remarkable proficiency in such reasoning tasks. This prompts a crucial question: have VLMs effectively tackled the CR challenge? We conjecture that existing CR benchmarks may not adequately push the boundaries of modern VLMs due to the reliance on an LLM-only negative text generation pipeline. Consequently, the negatives produced either appear as outliers from the natural language distribution learned by VLMs' LLM decoders or as improbable within the corresponding image context. To address these limitations, we introduce ConMe -- a compositional reasoning benchmark and a novel data generation pipeline leveraging VLMs to produce `hard CR Q&A'. Through a new concept of VLMs conversing with each other to collaboratively expose their weaknesses, our pipeline autonomously generates, evaluates, and selects challenging compositional reasoning questions, establishing a robust CR benchmark, also subsequently validated manually. Our benchmark provokes a noteworthy, up to 33%, decrease in CR performance compared to preceding benchmarks, reinstating the CR challenge even for state-of-the-art VLMs., Comment: The first three authors contributed equally

Low-rank adapters (LoRA) and their variants are popular parameter-efficient fine-tuning (PEFT) techniques that closely match full model fine-tune performance while requiring only a small number of additional parameters. These additional LoRA parameters are specific to the base model being adapted. When the base model needs to be deprecated and replaced with a new one, all the associated LoRA modules need to be re-trained. Such re-training requires access to the data used to train the LoRA for the original base model. This is especially problematic for commercial cloud applications where the LoRA modules and the base models are hosted by service providers who may not be allowed to host proprietary client task data. To address this challenge, we propose $\textit{Trans-LoRA}$ -- a novel method for lossless, nearly data-free transfer of LoRAs across base models. Our approach relies on synthetic data to transfer LoRA modules. Using large language models, we design a synthetic data generator to approximate the data-generating process of the $\textit{observed}$ task data subset. Training on the resulting synthetic dataset transfers LoRA modules to new models. We show the effectiveness of our approach using both LLama and Gemma model families. Our approach achieves lossless (mostly improved) LoRA transfer between models within and across different base model families, and even between different PEFT methods, on a wide variety of tasks.

Mixture-of-Experts (MoE) language models can reduce computational costs by 2-4$\times$ compared to dense models without sacrificing performance, making them more efficient in computation-bounded scenarios. However, MoE models generally require 2-4$\times$ times more parameters to achieve comparable performance to a dense model, which incurs larger GPU memory requirements and makes MoE models less efficient in I/O-bounded scenarios like autoregressive generation. In this work, we propose a hybrid dense training and sparse inference framework for MoE models (DS-MoE) which achieves strong computation and parameter efficiency by employing dense computation across all experts during training and sparse computation during inference. Our experiments on training LLMs demonstrate that our DS-MoE models are more parameter-efficient than standard sparse MoEs and are on par with dense models in terms of total parameter size and performance while being computationally cheaper (activating 30-40% of the model's parameters). Performance tests using vLLM show that our DS-MoE-6B model runs up to $1.86\times$ faster than similar dense models like Mistral-7B, and between $1.50\times$ and $1.71\times$ faster than comparable MoEs, such as DeepSeekMoE-16B and Qwen1.5-MoE-A2.7B.

Pre-training on massive video datasets has become essential to achieve high action recognition performance on smaller downstream datasets. However, most large-scale video datasets contain images of people and hence are accompanied with issues related to privacy, ethics, and data protection, often preventing them from being publicly shared for reproducible research. Existing work has attempted to alleviate these problems by blurring faces, downsampling videos, or training on synthetic data. On the other hand, analysis on the transferability of privacy-preserving pre-trained models to downstream tasks has been limited. In this work, we study this problem by first asking the question: can we pre-train models for human action recognition with data that does not include real humans? To this end, we present, for the first time, a benchmark that leverages real-world videos with humans removed and synthetic data containing virtual humans to pre-train a model. We then evaluate the transferability of the representation learned on this data to a diverse set of downstream action recognition benchmarks. Furthermore, we propose a novel pre-training strategy, called Privacy-Preserving MAE-Align, to effectively combine synthetic data and human-removed real data. Our approach outperforms previous baselines by up to 5% and closes the performance gap between human and no-human action recognition representations on downstream tasks, for both linear probing and fine-tuning. Our benchmark, code, and models are available at https://github.com/howardzh01/PPMA ., Comment: 19 pages, 7 figures, 2023 NeurIPS Datasets and Benchmarks Track

We explore the use of language as a perceptual representation for vision-and-language navigation (VLN), with a focus on low-data settings. Our approach uses off-the-shelf vision systems for image captioning and object detection to convert an agent's egocentric panoramic view at each time step into natural language descriptions. We then finetune a pretrained language model to select an action, based on the current view and the trajectory history, that would best fulfill the navigation instructions. In contrast to the standard setup which adapts a pretrained language model to work directly with continuous visual features from pretrained vision models, our approach instead uses (discrete) language as the perceptual representation. We explore several use cases of our language-based navigation (LangNav) approach on the R2R VLN benchmark: generating synthetic trajectories from a prompted language model (GPT-4) with which to finetune a smaller language model; domain transfer where we transfer a policy learned on one simulated environment (ALFRED) to another (more realistic) environment (R2R); and combining both vision- and language-based representations for VLN. Our approach is found to improve upon baselines that rely on visual features in settings where only a few expert trajectories (10-100) are available, demonstrating the potential of language as a perceptual representation for navigation.

The development of technologies for easily and automatically falsifying video has raised practical questions about people's ability to detect false information online. How vulnerable are people to deepfake videos? What technologies can be applied to boost their performance? Human susceptibility to deepfake videos is typically measured in laboratory settings, which do not reflect the challenges of real-world browsing. In typical browsing, deepfakes are rare, engagement with the video may be short, participants may be distracted, or the video streaming quality may be degraded. Here, we tested deepfake detection under these ecological viewing conditions, and found that detection was lowered in all cases. Principles from signal detection theory indicated that different viewing conditions affected different dimensions of detection performance. Overall, this suggests that the current literature underestimates people's susceptibility to deepfakes. Next, we examined how computer vision models might be integrated into users' decision process to increase accuracy and confidence during deepfake detection. We evaluated the effectiveness of communicating the model's prediction to the user by amplifying artifacts in fake videos. We found that artifact amplification was highly effective at making fake video distinguishable from real, in a manner that was robust across viewing conditions. Additionally, compared to a traditional text-based prompt, artifact amplification was more convincing: people accepted the model's suggestion more often, and reported higher final confidence in their model-supported decision, particularly for more challenging videos. Overall, this suggests that visual indicators that cause distortions on fake videos may be highly effective at mitigating the impact of falsified video., Comment: 8 pages, 4 figures

Large-scale pre-trained Vision & Language (VL) models have shown remarkable performance in many applications, enabling replacing a fixed set of supported classes with zero-shot open vocabulary reasoning over (almost arbitrary) natural language prompts. However, recent works have uncovered a fundamental weakness of these models. For example, their difficulty to understand Visual Language Concepts (VLC) that go 'beyond nouns' such as the meaning of non-object words (e.g., attributes, actions, relations, states, etc.), or difficulty in performing compositional reasoning such as understanding the significance of the order of the words in a sentence. In this work, we investigate to which extent purely synthetic data could be leveraged to teach these models to overcome such shortcomings without compromising their zero-shot capabilities. We contribute Synthetic Visual Concepts (SyViC) - a million-scale synthetic dataset and data generation codebase allowing to generate additional suitable data to improve VLC understanding and compositional reasoning of VL models. Additionally, we propose a general VL finetuning strategy for effectively leveraging SyViC towards achieving these improvements. Our extensive experiments and ablations on VL-Checklist, Winoground, and ARO benchmarks demonstrate that it is possible to adapt strong pre-trained VL models with synthetic data significantly enhancing their VLC understanding (e.g. by 9.9% on ARO and 4.3% on VL-Checklist) with under 1% drop in their zero-shot accuracy., Comment: Accepted to ICCV 2023. Project page: https://synthetic-vic.github.io/

Deepfakes pose a serious threat to digital well-being by fueling misinformation. As deepfakes get harder to recognize with the naked eye, human users become increasingly reliant on deepfake detection models to decide if a video is real or fake. Currently, models yield a prediction for a video's authenticity, but do not integrate a method for alerting a human user. We introduce a framework for amplifying artifacts in deepfake videos to make them more detectable by people. We propose a novel, semi-supervised Artifact Attention module, which is trained on human responses to create attention maps that highlight video artifacts. These maps make two contributions. First, they improve the performance of our deepfake detection classifier. Second, they allow us to generate novel "Deepfake Caricatures": transformations of the deepfake that exacerbate artifacts to improve human detection. In a user study, we demonstrate that Caricatures greatly increase human detection, across video presentation times and user engagement levels. Overall, we demonstrate the success of a human-centered approach to designing deepfake mitigation methods., Comment: 9 pages, 5 figures, 4 tables

We introduce Ego4D, a massive-scale egocentric video dataset and benchmark suite. It offers 3,670 hours of daily-life activity video spanning hundreds of scenarios (household, outdoor, workplace, leisure, etc.) captured by 931 unique camera wearers from 74 worldwide locations and 9 different countries. The approach to collection is designed to uphold rigorous privacy and ethics standards with consenting participants and robust de-identification procedures where relevant. Ego4D dramatically expands the volume of diverse egocentric video footage publicly available to the research community. Portions of the video are accompanied by audio, 3D meshes of the environment, eye gaze, stereo, and/or synchronized videos from multiple egocentric cameras at the same event. Furthermore, we present a host of new benchmark challenges centered around understanding the first-person visual experience in the past (querying an episodic memory), present (analyzing hand-object manipulation, audio-visual conversation, and social interactions), and future (forecasting activities). By publicly sharing this massive annotated dataset and benchmark suite, we aim to push the frontier of first-person perception. Project page: https://ego4d-data.org/, Comment: To appear in the Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2022. This version updates the baseline result numbers for the Hands and Objects benchmark (appendix)

Deep convolutional networks have recently achieved great success in video recognition, yet their practical realization remains a challenge due to the large amount of computational resources required to achieve robust recognition. Motivated by the effectiveness of quantization for boosting efficiency, in this paper, we propose a dynamic network quantization framework, that selects optimal precision for each frame conditioned on the input for efficient video recognition. Specifically, given a video clip, we train a very lightweight network in parallel with the recognition network, to produce a dynamic policy indicating which numerical precision to be used per frame in recognizing videos. We train both networks effectively using standard backpropagation with a loss to achieve both competitive performance and resource efficiency required for video recognition. Extensive experiments on four challenging diverse benchmark datasets demonstrate that our proposed approach provides significant savings in computation and memory usage while outperforming the existing state-of-the-art methods., Comment: ICCV 2021 Camera Ready Version

The self-attention-based model, transformer, is recently becoming the leading backbone in the field of computer vision. In spite of the impressive success made by transformers in a variety of vision tasks, it still suffers from heavy computation and intensive memory costs. To address this limitation, this paper presents an Interpretability-Aware REDundancy REDuction framework (IA-RED$^2$). We start by observing a large amount of redundant computation, mainly spent on uncorrelated input patches, and then introduce an interpretable module to dynamically and gracefully drop these redundant patches. This novel framework is then extended to a hierarchical structure, where uncorrelated tokens at different stages are gradually removed, resulting in a considerable shrinkage of computational cost. We include extensive experiments on both image and video tasks, where our method could deliver up to 1.4x speed-up for state-of-the-art models like DeiT and TimeSformer, by only sacrificing less than 0.7% accuracy. More importantly, contrary to other acceleration approaches, our method is inherently interpretable with substantial visual evidence, making vision transformer closer to a more human-understandable architecture while being lighter. We demonstrate that the interpretability that naturally emerged in our framework can outperform the raw attention learned by the original visual transformer, as well as those generated by off-the-shelf interpretation methods, with both qualitative and quantitative results. Project Page: http://people.csail.mit.edu/bpan/ia-red/., Comment: Accepted in NeurIPS 2021

Recent advances in representation learning have demonstrated an ability to represent information from different modalities such as video, text, and audio in a single high-level embedding vector. In this work we present a self-supervised learning framework that is able to learn a representation that captures finer levels of granularity across different modalities such as concepts or events represented by visual objects or spoken words. Our framework relies on a discretized embedding space created via vector quantization that is shared across different modalities. Beyond the shared embedding space, we propose a Cross-Modal Code Matching objective that forces the representations from different views (modalities) to have a similar distribution over the discrete embedding space such that cross-modal objects/actions localization can be performed without direct supervision. In our experiments we show that the proposed discretized multi-modal fine-grained representation (e.g., pixel/word/frame) can complement high-level summary representations (e.g., video/sentence/waveform) for improved performance on cross-modal retrieval tasks. We also observe that the discretized representation uses individual clusters to represent the same semantic concept across modalities., Comment: Preprint

Multi-modal learning, which focuses on utilizing various modalities to improve the performance of a model, is widely used in video recognition. While traditional multi-modal learning offers excellent recognition results, its computational expense limits its impact for many real-world applications. In this paper, we propose an adaptive multi-modal learning framework, called AdaMML, that selects on-the-fly the optimal modalities for each segment conditioned on the input for efficient video recognition. Specifically, given a video segment, a multi-modal policy network is used to decide what modalities should be used for processing by the recognition model, with the goal of improving both accuracy and efficiency. We efficiently train the policy network jointly with the recognition model using standard back-propagation. Extensive experiments on four challenging diverse datasets demonstrate that our proposed adaptive approach yields 35%-55% reduction in computation when compared to the traditional baseline that simply uses all the modalities irrespective of the input, while also achieving consistent improvements in accuracy over the state-of-the-art methods.

When people observe events, they are able to abstract key information and build concise summaries of what is happening. These summaries include contextual and semantic information describing the important high-level details (what, where, who and how) of the observed event and exclude background information that is deemed unimportant to the observer. With this in mind, the descriptions people generate for videos of different dynamic events can greatly improve our understanding of the key information of interest in each video. These descriptions can be captured in captions that provide expanded attributes for video labeling (e.g. actions/objects/scenes/sentiment/etc.) while allowing us to gain new insight into what people find important or necessary to summarize specific events. Existing caption datasets for video understanding are either small in scale or restricted to a specific domain. To address this, we present the Spoken Moments (S-MiT) dataset of 500k spoken captions each attributed to a unique short video depicting a broad range of different events. We collect our descriptions using audio recordings to ensure that they remain as natural and concise as possible while allowing us to scale the size of a large classification dataset. In order to utilize our proposed dataset, we present a novel Adaptive Mean Margin (AMM) approach to contrastive learning and evaluate our models on video/caption retrieval on multiple datasets. We show that our AMM approach consistently improves our results and that models trained on our Spoken Moments dataset generalize better than those trained on other video-caption datasets., Comment: To appear at CVPR 2021

The pixels in an image, and the objects, scenes, and actions that they compose, determine whether an image will be memorable or forgettable. While memorability varies by image, it is largely independent of an individual observer. Observer independence is what makes memorability an image-computable measure of information, and eligible for automatic prediction. In this chapter, we zoom into memorability with a computational lens, detailing the state-of-the-art algorithms that accurately predict image memorability relative to human behavioral data, using image features at different scales from raw pixels to semantic labels. We discuss the design of algorithms and visualizations for face, object, and scene memorability, as well as algorithms that generalize beyond static scenes to actions and videos. We cover the state-of-the-art deep learning approaches that are the current front runners in the memorability prediction space. Beyond prediction, we show how recent A.I. approaches can be used to create and modify visual memorability. Finally, we preview the computational applications that memorability can power, from filtering visual streams to enhancing augmented reality interfaces.

Quantizing deep networks with adaptive bit-widths is a promising technique for efficient inference across many devices and resource constraints. In contrast to static methods that repeat the quantization process and train different models for different constraints, adaptive quantization enables us to flexibly adjust the bit-widths of a single deep network during inference for instant adaptation in different scenarios. While existing research shows encouraging results on common image classification benchmarks, this paper investigates how to train such adaptive networks more effectively. Specifically, we present two novel techniques for quantizing deep neural networks with adaptive bit-widths of weights and activations. First, we propose a collaborative strategy to choose a high-precision teacher for transferring knowledge to the low-precision student while jointly optimizing the model with all bit-widths. Second, to effectively transfer knowledge, we develop a dynamic block swapping method by randomly replacing the blocks in the lower-precision student network with the corresponding blocks in the higher-precision teacher network. Extensive experiments on multiple image classification datasets including video classification benchmarks for the first time, well demonstrate the efficacy of our approach over state-of-the-art methods.

Performing inference on deep learning models for videos remains a challenge due to the large amount of computational resources required to achieve robust recognition. An inherent property of real-world videos is the high correlation of information across frames which can translate into redundancy in either temporal or spatial feature maps of the models, or both. The type of redundant features depends on the dynamics and type of events in the video: static videos have more temporal redundancy while videos focusing on objects tend to have more channel redundancy. Here we present a redundancy reduction framework, termed VA-RED$^2$, which is input-dependent. Specifically, our VA-RED$^2$ framework uses an input-dependent policy to decide how many features need to be computed for temporal and channel dimensions. To keep the capacity of the original model, after fully computing the necessary features, we reconstruct the remaining redundant features from those using cheap linear operations. We learn the adaptive policy jointly with the network weights in a differentiable way with a shared-weight mechanism, making it highly efficient. Extensive experiments on multiple video datasets and different visual tasks show that our framework achieves $20\% - 40\%$ reduction in computation (FLOPs) when compared to state-of-the-art methods without any performance loss. Project page: http://people.csail.mit.edu/bpan/va-red/., Comment: Accepted in ICLR 2021

Temporal modelling is the key for efficient video action recognition. While understanding temporal information can improve recognition accuracy for dynamic actions, removing temporal redundancy and reusing past features can significantly save computation leading to efficient action recognition. In this paper, we introduce an adaptive temporal fusion network, called AdaFuse, that dynamically fuses channels from current and past feature maps for strong temporal modelling. Specifically, the necessary information from the historical convolution feature maps is fused with current pruned feature maps with the goal of improving both recognition accuracy and efficiency. In addition, we use a skipping operation to further reduce the computation cost of action recognition. Extensive experiments on Something V1 & V2, Jester and Mini-Kinetics show that our approach can achieve about 40% computation savings with comparable accuracy to state-of-the-art methods. The project page can be found at https://mengyuest.github.io/AdaFuse/, Comment: Accepted to ICLR2021

In recent years, a number of approaches based on 2D or 3D convolutional neural networks (CNN) have emerged for video action recognition, achieving state-of-the-art results on several large-scale benchmark datasets. In this paper, we carry out in-depth comparative analysis to better understand the differences between these approaches and the progress made by them. To this end, we develop an unified framework for both 2D-CNN and 3D-CNN action models, which enables us to remove bells and whistles and provides a common ground for fair comparison. We then conduct an effort towards a large-scale analysis involving over 300 action recognition models. Our comprehensive analysis reveals that a) a significant leap is made in efficiency for action recognition, but not in accuracy; b) 2D-CNN and 3D-CNN models behave similarly in terms of spatio-temporal representation abilities and transferability. Our codes are available at https://github.com/IBM/action-recognition-pytorch., Comment: CVPR 2021 camera-ready version. Codes and models are available on https://github.com/IBM/action-recognition-pytorch

A key capability of an intelligent system is deciding when events from past experience must be remembered and when they can be forgotten. Towards this goal, we develop a predictive model of human visual event memory and how those memories decay over time. We introduce Memento10k, a new, dynamic video memorability dataset containing human annotations at different viewing delays. Based on our findings we propose a new mathematical formulation of memorability decay, resulting in a model that is able to produce the first quantitative estimation of how a video decays in memory over time. In contrast with previous work, our model can predict the probability that a video will be remembered at an arbitrary delay. Importantly, our approach combines visual and semantic information (in the form of textual captions) to fully represent the meaning of events. Our experiments on two video memorability benchmarks, including Memento10k, show that our model significantly improves upon the best prior approach (by 12% on average)., Comment: European Conference on Computer Vision

Identifying common patterns among events is a key ability in human and machine perception, as it underlies intelligent decision making. We propose an approach for learning semantic relational set abstractions on videos, inspired by human learning. We combine visual features with natural language supervision to generate high-level representations of similarities across a set of videos. This allows our model to perform cognitive tasks such as set abstraction (which general concept is in common among a set of videos?), set completion (which new video goes well with the set?), and odd one out detection (which video does not belong to the set?). Experiments on two video benchmarks, Kinetics and Multi-Moments in Time, show that robust and versatile representations emerge when learning to recognize commonalities among sets. We compare our model to several baseline algorithms and show that significant improvements result from explicitly learning relational abstractions with semantic supervision., Comment: European Conference on Computer Vision (ECCV) 2020, accepted

Action recognition is an open and challenging problem in computer vision. While current state-of-the-art models offer excellent recognition results, their computational expense limits their impact for many real-world applications. In this paper, we propose a novel approach, called AR-Net (Adaptive Resolution Network), that selects on-the-fly the optimal resolution for each frame conditioned on the input for efficient action recognition in long untrimmed videos. Specifically, given a video frame, a policy network is used to decide what input resolution should be used for processing by the action recognition model, with the goal of improving both accuracy and efficiency. We efficiently train the policy network jointly with the recognition model using standard back-propagation. Extensive experiments on several challenging action recognition benchmark datasets well demonstrate the efficacy of our proposed approach over state-of-the-art methods. The project page can be found at https://mengyuest.github.io/AR-Net

Videos capture events that typically contain multiple sequential, and simultaneous, actions even in the span of only a few seconds. However, most large-scale datasets built to train models for action recognition in video only provide a single label per video. Consequently, models can be incorrectly penalized for classifying actions that exist in the videos but are not explicitly labeled and do not learn the full spectrum of information present in each video in training. Towards this goal, we present the Multi-Moments in Time dataset (M-MiT) which includes over two million action labels for over one million three second videos. This multi-label dataset introduces novel challenges on how to train and analyze models for multi-action detection. Here, we present baseline results for multi-action recognition using loss functions adapted for long tail multi-label learning, provide improved methods for visualizing and interpreting models trained for multi-label action detection and show the strength of transferring models trained on M-MiT to smaller datasets.

Objects are entities we act upon, where the functionality of an object is determined by how we interact with it. In this work we propose a Dual Attention Network model which reasons about human-object interactions. The dual-attentional framework weights the important features for objects and actions respectively. As a result, the recognition of objects and actions mutually benefit each other. The proposed model shows competitive classification performance on the human-object interaction dataset Something-Something. Besides, it can perform weak spatiotemporal localization and affordance segmentation, despite being trained only with video-level labels. The model not only finds when an action is happening and which object is being manipulated, but also identifies which part of the object is being interacted with. Project page: \url{https://dual-attention-network.github.io/}., Comment: ICCV 2019

We introduce a framework that uses Generative Adversarial Networks (GANs) to study cognitive properties like memorability, aesthetics, and emotional valence. These attributes are of interest because we do not have a concrete visual definition of what they entail. What does it look like for a dog to be more or less memorable? GANs allow us to generate a manifold of natural-looking images with fine-grained differences in their visual attributes. By navigating this manifold in directions that increase memorability, we can visualize what it looks like for a particular generated image to become more or less memorable. The resulting ``visual definitions" surface image properties (like ``object size") that may underlie memorability. Through behavioral experiments, we verify that our method indeed discovers image manipulations that causally affect human memory performance. We further demonstrate that the same framework can be used to analyze image aesthetics and emotional valence. Visit the GANalyze website at http://ganalyze.csail.mit.edu/., Comment: 17 pages, 15 figures

In the last decade, artificial intelligence (AI) models inspired by the brain have made unprecedented progress in performing real-world perceptual tasks like object classification and speech recognition. Recently, researchers of natural intelligence have begun using those AI models to explore how the brain performs such tasks. These developments suggest that future progress will benefit from increased interaction between disciplines. Here we introduce the Algonauts Project as a structured and quantitative communication channel for interdisciplinary interaction between natural and artificial intelligence researchers. The project's core is an open challenge with a quantitative benchmark whose goal is to account for brain data through computational models. This project has the potential to provide better models of natural intelligence and to gather findings that advance AI. The 2019 Algonauts Project focuses on benchmarking computational models predicting human brain activity when people look at pictures of objects. The 2019 edition of the Algonauts Project is available online: http://algonauts.csail.mit.edu/., Comment: 4 pages, 2 figures

Widely used in news, business, and educational media, infographics are handcrafted to effectively communicate messages about complex and often abstract topics including `ways to conserve the environment' and `understanding the financial crisis'. Composed of stylistically and semantically diverse visual and textual elements, infographics pose new challenges for computer vision. While automatic text extraction works well on infographics, computer vision approaches trained on natural images fail to identify the stand-alone visual elements in infographics, or `icons'. To bridge this representation gap, we propose a synthetic data generation strategy: we augment background patches in infographics from our Visually29K dataset with Internet-scraped icons which we use as training data for an icon proposal mechanism. On a test set of 1K annotated infographics, icons are located with 38% precision and 34% recall (the best model trained with natural images achieves 14% precision and 7% recall). Combining our icon proposals with icon classification and text extraction, we present a multi-modal summarization application. Our application takes an infographic as input and automatically produces text tags and visual hashtags that are textually and visually representative of the infographic's topics respectively.

We present the Moments in Time Dataset, a large-scale human-annotated collection of one million short videos corresponding to dynamic events unfolding within three seconds. Modeling the spatial-audio-temporal dynamics even for actions occurring in 3 second videos poses many challenges: meaningful events do not include only people, but also objects, animals, and natural phenomena; visual and auditory events can be symmetrical in time ("opening" is "closing" in reverse), and either transient or sustained. We describe the annotation process of our dataset (each video is tagged with one action or activity label among 339 different classes), analyze its scale and diversity in comparison to other large-scale video datasets for action recognition, and report results of several baseline models addressing separately, and jointly, three modalities: spatial, temporal and auditory. The Moments in Time dataset, designed to have a large coverage and diversity of events in both visual and auditory modalities, can serve as a new challenge to develop models that scale to the level of complexity and abstract reasoning that a human processes on a daily basis.

Temporal relational reasoning, the ability to link meaningful transformations of objects or entities over time, is a fundamental property of intelligent species. In this paper, we introduce an effective and interpretable network module, the Temporal Relation Network (TRN), designed to learn and reason about temporal dependencies between video frames at multiple time scales. We evaluate TRN-equipped networks on activity recognition tasks using three recent video datasets - Something-Something, Jester, and Charades - which fundamentally depend on temporal relational reasoning. Our results demonstrate that the proposed TRN gives convolutional neural networks a remarkable capacity to discover temporal relations in videos. Through only sparsely sampled video frames, TRN-equipped networks can accurately predict human-object interactions in the Something-Something dataset and identify various human gestures on the Jester dataset with very competitive performance. TRN-equipped networks also outperform two-stream networks and 3D convolution networks in recognizing daily activities in the Charades dataset. Further analyses show that the models learn intuitive and interpretable visual common sense knowledge in videos., Comment: camera-ready version for ECCV'18

The success of recent deep convolutional neural networks (CNNs) depends on learning hidden representations that can summarize the important factors of variation behind the data. However, CNNs often criticized as being black boxes that lack interpretability, since they have millions of unexplained model parameters. In this work, we describe Network Dissection, a method that interprets networks by providing labels for the units of their deep visual representations. The proposed method quantifies the interpretability of CNN representations by evaluating the alignment between individual hidden units and a set of visual semantic concepts. By identifying the best alignments, units are given human interpretable labels across a range of objects, parts, scenes, textures, materials, and colors. The method reveals that deep representations are more transparent and interpretable than expected: we find that representations are significantly more interpretable than they would be under a random equivalently powerful basis. We apply the method to interpret and compare the latent representations of various network architectures trained to solve different supervised and self-supervised training tasks. We then examine factors affecting the network interpretability such as the number of the training iterations, regularizations, different initializations, and the network depth and width. Finally we show that the interpreted units can be used to provide explicit explanations of a prediction given by a CNN for an image. Our results highlight that interpretability is an important property of deep neural networks that provides new insights into their hierarchical structure., Comment: *B. Zhou and D. Bau contributed equally to this work. 15 pages, 27 figures

We introduce the problem of visual hashtag discovery for infographics: extracting visual elements from an infographic that are diagnostic of its topic. Given an infographic as input, our computational approach automatically outputs textual and visual elements predicted to be representative of the infographic content. Concretely, from a curated dataset of 29K large infographic images sampled across 26 categories and 391 tags, we present an automated two step approach. First, we extract the text from an infographic and use it to predict text tags indicative of the infographic content. And second, we use these predicted text tags as a supervisory signal to localize the most diagnostic visual elements from within the infographic i.e. visual hashtags. We report performances on a categorization and multi-label tag prediction problem and compare our proposed visual hashtags to human annotations.

We propose a general framework called Network Dissection for quantifying the interpretability of latent representations of CNNs by evaluating the alignment between individual hidden units and a set of semantic concepts. Given any CNN model, the proposed method draws on a broad data set of visual concepts to score the semantics of hidden units at each intermediate convolutional layer. The units with semantics are given labels across a range of objects, parts, scenes, textures, materials, and colors. We use the proposed method to test the hypothesis that interpretability of units is equivalent to random linear combinations of units, then we apply our method to compare the latent representations of various networks when trained to solve different supervised and self-supervised training tasks. We further analyze the effect of training iterations, compare networks trained with different initializations, examine the impact of network depth and width, and measure the effect of dropout and batch normalization on the interpretability of deep visual representations. We demonstrate that the proposed method can shed light on characteristics of CNN models and training methods that go beyond measurements of their discriminative power., Comment: First two authors contributed equally. Oral presentation at CVPR 2017

In this paper, we present BubbleView, an alternative methodology for eye tracking using discrete mouse clicks to measure which information people consciously choose to examine. BubbleView is a mouse-contingent, moving-window interface in which participants are presented with a series of blurred images and click to reveal "bubbles" - small, circular areas of the image at original resolution, similar to having a confined area of focus like the eye fovea. Across 10 experiments with 28 different parameter combinations, we evaluated BubbleView on a variety of image types: information visualizations, natural images, static webpages, and graphic designs, and compared the clicks to eye fixations collected with eye-trackers in controlled lab settings. We found that BubbleView clicks can both (i) successfully approximate eye fixations on different images, and (ii) be used to rank image and design elements by importance. BubbleView is designed to collect clicks on static images, and works best for defined tasks such as describing the content of an information visualization or measuring image importance. BubbleView data is cleaner and more consistent than related methodologies that use continuous mouse movements. Our analyses validate the use of mouse-contingent, moving-window methodologies as approximating eye fixations for different image and task types.

The rise of multi-million-item dataset initiatives has enabled data-hungry machine learning algorithms to reach near-human semantic classification at tasks such as object and scene recognition. Here we describe the Places Database, a repository of 10 million scene photographs, labeled with scene semantic categories and attributes, comprising a quasi-exhaustive list of the types of environments encountered in the world. Using state of the art Convolutional Neural Networks, we provide impressive baseline performances at scene classification. With its high-coverage and high-diversity of exemplars, the Places Database offers an ecosystem to guide future progress on currently intractable visual recognition problems.

The next grand challenges for society and science are in the brain sciences. A collection of 60+ scientists from around the world, together with 10+ observers from national, private, and foundations, spent two days together discussing the top challenges that we could solve as a global community in the next decade. We eventually settled on three challenges, spanning anatomy, physiology, and medicine. Addressing all three challenges requires novel computational infrastructure. The group proposed the advent of The International Brain Station (TIBS), to address these challenges, and launch brain sciences to the next level of understanding., Comment: 6 pages

How best to evaluate a saliency model's ability to predict where humans look in images is an open research question. The choice of evaluation metric depends on how saliency is defined and how the ground truth is represented. Metrics differ in how they rank saliency models, and this results from how false positives and false negatives are treated, whether viewing biases are accounted for, whether spatial deviations are factored in, and how the saliency maps are pre-processed. In this paper, we provide an analysis of 8 different evaluation metrics and their properties. With the help of systematic experiments and visualizations of metric computations, we add interpretability to saliency scores and more transparency to the evaluation of saliency models. Building off the differences in metric properties and behaviors, we make recommendations for metric selections under specific assumptions and for specific applications.

The complex multi-stage architecture of cortical visual pathways provides the neural basis for efficient visual object recognition in humans. However, the stage-wise computations therein remain poorly understood. Here, we compared temporal (magnetoencephalography) and spatial (functional MRI) visual brain representations with representations in an artificial deep neural network (DNN) tuned to the statistics of real-world visual recognition. We showed that the DNN captured the stages of human visual processing in both time and space from early visual areas towards the dorsal and ventral streams. Further investigation of crucial DNN parameters revealed that while model architecture was important, training on real-world categorization was necessary to enforce spatio-temporal hierarchical relationships with the brain. Together our results provide an algorithmically informed view on the spatio-temporal dynamics of visual object recognition in the human visual brain., Comment: 15 pages, 6 figures

In this work, we revisit the global average pooling layer proposed in [13], and shed light on how it explicitly enables the convolutional neural network to have remarkable localization ability despite being trained on image-level labels. While this technique was previously proposed as a means for regularizing training, we find that it actually builds a generic localizable deep representation that can be applied to a variety of tasks. Despite the apparent simplicity of global average pooling, we are able to achieve 37.1% top-5 error for object localization on ILSVRC 2014, which is remarkably close to the 34.2% top-5 error achieved by a fully supervised CNN approach. We demonstrate that our network is able to localize the discriminative image regions on a variety of tasks despite not being trained for them

With the success of new computational architectures for visual processing, such as convolutional neural networks (CNN) and access to image databases with millions of labeled examples (e.g., ImageNet, Places), the state of the art in computer vision is advancing rapidly. One important factor for continued progress is to understand the representations that are learned by the inner layers of these deep architectures. Here we show that object detectors emerge from training CNNs to perform scene classification. As scenes are composed of objects, the CNN for scene classification automatically discovers meaningful objects detectors, representative of the learned scene categories. With object detectors emerging as a result of learning to recognize scenes, our work demonstrates that the same network can perform both scene recognition and object localization in a single forward-pass, without ever having been explicitly taught the notion of objects., Comment: 12 pages, ICLR 2015 conference paper

Although the human visual system can recognize many concepts under challenging conditions, it still has some biases. In this paper, we investigate whether we can extract these biases and transfer them into a machine recognition system. We introduce a novel method that, inspired by well-known tools in human psychophysics, estimates the biases that the human visual system might use for recognition, but in computer vision feature spaces. Our experiments are surprising, and suggest that classifiers from the human visual system can be transferred into a machine with some success. Since these classifiers seem to capture favorable biases in the human visual system, we further present an SVM formulation that constrains the orientation of the SVM hyperplane to agree with the bias from human visual system. Our results suggest that transferring this human bias into machines may help object recognition systems generalize across datasets and perform better when very little training data is available., Comment: To appear at NIPS 2015