Search

Indoor 3D object detection is an essential task in single image scene understanding, impacting spatial cognition fundamentally in visual reasoning. Existing works on 3D object detection from a single image either pursue this goal through independent predictions of each object or implicitly reason over all possible objects, failing to harness relational geometric information between objects. To address this problem, we propose a sparse graph-based pipeline named Explicit3D based on object geometry and semantics features. Taking the efficiency into consideration, we further define a relatedness score and design a novel dynamic pruning method via group sampling for sparse scene graph generation and updating. Furthermore, our Explicit3D introduces homogeneous matrices and defines new relative loss and corner loss to model the spatial difference between target pairs explicitly. Instead of using ground-truth labels as direct supervision, our relative and corner loss are derived from homogeneous transforms, which renders the model to learn the geometric consistency between objects. The experimental results on the SUN RGB-D dataset demonstrate that our Explicit3D achieves better performance balance than the-state-of-the-art. • We propose a sparse graph-based pipeline for single image 3D indoor scene detection. • A novel relatedness score guides dynamic pruning for sparse scene graph generation. • Homogeneous matrices are used to explicitly represent relative spatial relationships. • A novel relative loss boosts spatial offset prediction and overall object detection. [ABSTRACT FROM AUTHOR]

In fact, the use of statistics, including the local and global summary statistics, statistical models and statistical machine learning techniques, becomes more and more popular in each constituent module of VQA. In the first paper, the authors study the influence of image content type on image quality assessment (IQA). In the fifth paper, the authors propose a low complexity wavelet based image quality evaluator. [Extracted from the article]

Increasing inter-class variance and shrinking intra-class distance are two main concerns and efforts in face recognition. In this paper, we propose a new loss function termed inter-class angular margin (IAM) loss aiming to enlarge the inter-class variance. Instead of restricting the inter-class margin to be a constant in existing methods, our IAM loss adaptively penalizes smaller inter-class angles more heavily and successfully makes the angular margin between classes larger, which can significantly enhance the discrimination of facial features. The IAM loss can be readily introduced as a regularization term for the widely-used Softmax loss and its recent variants to further improve their performances. We also analyze and verify the appropriate range of the regularization hyper-parameter from the perspective of backpropagation. For illustrative purposes, our model is trained on CASIA-WebFace and tested on the LFW, CFP, YTF and MegaFace datasets; the experimental results show that the IAM loss is quite effective to improve state-of-the-art algorithms. • We propose an IAM loss to increase the inter-class margin adaptively. • As a regularization term, IAM loss can be applied to current state-of-the-art methods and increase the recognition accuracy further. • Experiments on LFW, YTF, CFP and MegaFace can verify the effectiveness of our IAM loss. [ABSTRACT FROM AUTHOR]

Image quality assessment is usually achieved by pooling local quality scores. However, commonly used pooling strategies, based on simple sample statistics, are not always sensitive to distortions. In this short communication, we propose a novel perspective of pooling: reliable pooling through statistical hypothesis testing, which enables effective detection of subtle changes of population parameters when the underlying distribution of local quality scores is affected by distortions. To illustrate the significance of this novel perspective, we design a new pooling strategy utilising simple one-sided one-sample t -test. The experiments on benchmark databases show the reliability of hypothesis testing-based pooling, compared with state-of-the-art pooling strategies. [ABSTRACT FROM AUTHOR]

Face images in real Closed-Circuit Television (CCTV) are usually with low resolution, which remarkably deteriorates the performance of existing face recognition algorithms and hinders the application of face recognition. The main technical focus of this issue, matching between high-resolution (HR) and low-resolution (LR) face images has attracted significant attention. In order to better address this problem, we propose a Classifier Shared Deep Network with Multi-Hierarchy Loss (CS-MHL-Net) for low-resolution face recognition (LRFR) in this paper. Firstly, considering that contrastive loss and its variants are not conducive to the convergence of network and the reduction of discrepancy, a shared classifier between HR and LR face images is proposed to further narrow the domain gap between HR and LR by sharing the corresponding weights which can be seen as the class center. Secondly, to fully exploit intermediate features and loss constraints, we embed multi-hierarchy loss into intermediate layers, with the target of reducing the distances between HR and LR intermediate features after max pooling and avoiding the decreasing of accuracy caused by over-utilization of intermediate features. Experimental results on LFW and SCface demonstrate the effectiveness and superiority of the proposed method. • We share the classifier of LR and HR network, which can be seen as center-based method, and enforce the losses between HR and LR intermediate layers to narrow the discrimination of HR domain and LR domain. [ABSTRACT FROM AUTHOR]

Low-light image enhancement has attracted great attention due to its capability in providing visible content and textures for video security and forensics. Unfortunately, conventional Retinex-based methods hardly pay attention to the complementarity of illumination and reflection. Inspired by the observation, we propose a new low-light image enhancement framework which relights the restored low-light image based on complementarity of illumination and reflection, named R2Net. The R2Net basically consists of three parts: image decomposition (Decom Net), reflection restoration (Restore Net), and illumination enhancement (Relight Net). In the Restore Net, we propose a mixed twofold attention (MTFA) module with linear complexity, introducing illumination information to model the mutual relationship between illumination and reflection. In MTFA, in order to capture richer information, we first map the inputs into different feature spaces, followed by combining them in different orders to obtain multiple sets of enhanced features. Accordingly, we design a differential enhanced fusion module (DEFM) to mix the multiple features. Ablation studies prove that the MTFA module significantly improves the performance of the restoration network. Finally, we propose a new illumination enhancement network (Relight Net), in which we introduce reflection to generate global information and define a luminance factor to tune the exposure level of output images, making our model more robust and flexible. Experiments show that our proposed method outperforms state-of-the-art methods in both quantitative comparison and visual perception. • A R 2 Net is proposed based on complementarity of illumination and reflection. • A mixed twofold attention builds interaction between illumination and reflection. • A flexible network that can control the exposure level of enhanced images. [ABSTRACT FROM AUTHOR]