论文检索

Yaniv Benny, Lior Wolf

Foreground Segmentation, and Fine-Grained Clustering","We present a method for simultaneously learning, in an unsupervised manner, (i) a conditional image generator, (ii) foreground extraction and segmentation, (iii) clustering into a two-level class hierarchy, and (iv) object removal and background completion, all done without any use of annotation. The method combines a generative adversarial network and a variational autoencoder, with multiple encoders, generators and discriminators, and benefits from solving all tasks at once. The input to the training scheme is a varied collection of unlabeled images from the same domain, as well as a set of background images without a foreground object. In addition, the image generator can mix the background from one image, with a foreground that is conditioned either on that of a second image or on the index of a desired cluster. The method obtains state of the art results in comparison to the literature methods, when compared to the current state of the art in each of the tasks.

Kshitij Dwivedi, Jiahui Huang, Radoslaw Martin Cichy, Gemma Roig

In this paper, we tackle an open research question in transfer learning, which is selecting a model initialization to achieve high performance on a new task, given several pre-trained models. We propose a new highly efficient and accurate approach based on duality diagram similarity (DDS) between deep neural networks (DNNs). DDS is a generic framework to represent and compare data of different feature dimensions. We validate our approach on the Taskonomy dataset by measuring the correspondence between actual transfer learning performance rankings on 17 taskonomy tasks and predicted rankings. Computing DDS based ranking for $17 imes17$ transfers requires less than 2 minutes and shows a high correlation ($0.86$) with actual transfer learning rankings, outperforming state-of-the-art methods by a large margin ($10\%$) on the Taskonomy benchmark. We also demonstrate the robustness of our model selection approach to a new task, namely Pascal VOC semantic segmentation. Additionally, we show that our method can be applied to select the best layer locations within a DNN for transfer learning on 2D, 3D and semantic tasks on NYUv2 and Pascal VOC datasets.

Weizeng Lu, Xi Jia, Weicheng Xie, Linlin Shen, Yicong Zhou, Jinming Duan

We propose a geometry constrained network, termed GCNet, for weakly supervised object localization (WSOL). GC-Net consists of three modules: a detector, a generator and a classifier. The detector predicts the object location defined by a set of coefficients describing a geometric shape (i.e. ellipse or rectangle), which is geometrically constrained by the mask produced by the generator. The classifier takes the resulting masked images as input and performs two complementary classification tasks for the object and background. To make the mask more compact and more complete, we propose a novel multi-task loss function that takes into account area of the geometric shape, the categorical cross entropy and the negative entropy. In contrast to previous approaches, GC-Net is trained end-to-end and predict object location without any post-processing (e.g. thresholding) that may require additional tuning. Extensive experiments on the CUB-200-2011 and ILSVRC2012 datasets show that GC-Net outperforms state-of-the-art methods by a large margin. Our source code is available at https://github.com/lwzeng/GC-Net.

Christopher Zach, Huu Le

Optimization problems with an auxiliary latent variable structure in addition to the main model parameters occur frequently in computer vision and machine learning. The additional latent variables make the underlying optimization task expensive, either in terms of memory (by maintaining the latent variables), or in terms of runtime (repeated exact inference of latent variables). We aim to remove the need to maintain the latent variables and propose two formally justified methods, that dynamically adapt the required accuracy of latent variable inference. These methods have applications in large scale robust estimation and in learning the parameters of an energy-based model from labeled data.

Hai Victor Habi, Roy H. Jennings, Arnon Netzer

Recent work in network quantization produced state-of-the-art results using mixed precision quantization. An imperative requirement for many efficient edge device hardware implementations is that their quantizers are uniform and with power-of-two thresholds. In this work, we introduce the Hardware Friendly Mixed Precision Quantization Block (HMQ) in order to meet this requirement. The HMQ is a mixed precision quantization block that repurposes the Gumbel-Softmax estimator into a smooth estimator of a pair of quantization parameters, namely, bit-width and threshold. HMQs use this to search over a finite space of quantization schemes. Empirically, we apply HMQs to quantize classification models trained on CIFAR10 and ImageNet. For ImageNet, we quantize four different architectures and show that, in spite of the added restrictions to our quantization scheme, we achieve competitive and, in some cases, state-of-the-art results.

Juseung Yun, Byungjoo Kim, Junmo Kim

Although convolutional neural networks perform extremely well for numerous computer vision tasks, a considerably large amount of labeled data is required to ensure a good outcome. Data labeling is labor-intensive, and in some cases, the labeling budget may be limited. Active learning is a technique that can reduce the labeling required. With this technique, the neural network selects on its own the unlabeled data most helpful for learning, and then requests the human annotator for the labels. Most existing active learning methods have focused on acquisition functions for an effective selection of the informative samples. However, in this paper, we focus on the data-incremental nature of active learning, and propose a method for properly tuning the weight decay as the amount of data increases. We also demonstrate that the performance can be improved by knowledge distillation using a low-performance teacher model trained from the previous acquisition step. In addition, we present a novel perspective of the weight decay, which provides a regularization effect by limiting the number of effective parameters and channels in the convolutional filter. We validate our methods on the MNIST, CIFAR-10, and CIFAR-100 datasets using convolutional neural networks of various sizes.

Ke Mei, Chuang Zhu, Jiaqi Zou, Shanghang Zhang

The divergence between labeled training data and unlabeled testing data is a significant challenge for recent deep learning models. Unsupervised domain adaptation (UDA) attempts to solve such a problem. Recent works show that self-training is a powerful approach to UDA. However, existing methods have difficulty in balancing scalability and performance. In this paper, we propose an instance adaptive self-training framework for UDA on the task of semantic segmentation. To effectively improve the quality of pseudo-labels, we develop a novel pseudo-label generation strategy with an instance adaptive selector. Besides, we propose the region-guided regularization to smooth the pseudo-label region and sharpen the non-pseudo-label region. Our method is so concise and efficient that it is easy to be generalized to other unsupervised domain adaptation methods. Experiments on `GTA5 to Cityscapes' and `SYNTHIA to Cityscapes' demonstrate the superior performance of our approach compared with the state-of-the-art methods.

Nikolas Adaloglou, Nicholas Vretos, Petros Daras

In this paper, a novel multi-view methodology for graph-based neural networks is proposed. A systematic and methodological adaptation of the key concepts of classical deep learning methods such as convolution, pooling and multi-view architectures is developed for the context of non-Euclidean manifolds. The aim of the proposed work is to present a novel multi-view graph convolution layer, as well as a new view pooling layer making use of: a) a new hybrid Laplacian that is adjusted based on feature distance metric learning, b) multiple trainable representations of a feature matrix of a graph, using trainable distance matrices, adapting the notion of views to graphs and c) a multi-view graph aggregation scheme called graph view pooling, in order to synthesise information from the multiple generated ""views"". The aforementioned layers are used in an end-to-end graph neural network architecture for graph classification and show competitive results to other state-of-the-art methods.

Philippe Weinzaepfel, Romain Brégier, Hadrien Combaluzier, Vincent Leroy, Grégory Rogez

We introduce DOPE, the first method to detect and estimate whole-body 3D human poses, including bodies, hands and faces, in the wild. Achieving this level of details is key for a number of applications that require understanding the interactions of the people with each other or with the environment. The main challenge is the lack of in-the-wild data with labeled whole-body 3D poses. In previous work, training data has been annotated or generated for simpler tasks focusing on bodies, hands or faces separately. In this work, we propose to take advantage of these datasets to train independent experts for each part, namely a body, a hand and a face expert, and distill their knowledge into a single deep network designed for whole-body 2D-3D pose detection. In practice, given a training image with partial or no annotation, each part expert detects its subset of keypoints in 2D and 3D and the resulting estimations are combined to obtain whole-body pseudo ground-truth poses. A distillation loss encourages the whole-body predictions to mimic the experts' outputs. Our results show that this approach significantly outperforms the same whole-body model trained without distillation while staying close to the performance of the experts. Importantly, DOPE is computationally less demanding than the ensemble of experts and can achieve real-time performance. Test code and models are available at https://europe.naverlabs.com/research/computer-vision/dope.

Wallace Lira, Johannes Merz, Daniel Ritchie, Daniel Cohen-Or, Hao Zhang

We introduce GANHopper, an unsupervised image-to-image translation network that transforms images gradually between two domains, through multiple hops. Instead of executing translation directly, we steer the translation by requiring the network to produce in-between images that resemble weighted hybrids between images from the input domains. Our network is trained on unpaired images from the two domains only, without any in-between images. All hops are produced using a single generator along each direction. In addition to the standard cycle-consistency and adversarial losses, we introduce a new hybrid discriminator, which is trained to classify the intermediate images produced by the generator as weighted hybrids, with weights based on a predetermined hop count. We also add a smoothness term to constrain the magnitude of each hop, further regularizing the translation. Compared to previous methods, GANHopper excels at image translations involving domain-specific image features and geometric variations while also preserving non-domain-specific features such as general color schemes

Liyi Chen, Weiwei Wu, Chenchen Fu, Xiao Han, Yuntao Zhang

Weakly supervised semantic segmentation with image-level labels has attracted a lot of attention recently because these labels are already available in most datasets. To obtain semantic segmentation under weak supervision, this paper presents a simple yet effective approach based on the idea of explicitly exploring object boundaries from training images to keep coincidence of segmentation and boundaries. Specifically, we synthesize boundary annotations by exploiting coarse localization maps obtained from CNN classifier, and use annotations to train the proposed network called BENet which further excavates more object boundaries to provide constraints for segmentation. Finally generated pseudo annotations of training images are used to supervise an off-the-shelf segmentation network. We evaluate the proposed method on PASCAL VOC 2012 benchmark and the final results achieve 65.7% and 66.6% mIoU scores on val and test sets respectively, which outperforms previous methods trained under image-level supervision.

Christian Reimers, Jakob Runge, Joachim Denzler

Deep neural networks are tremendously successful in many applications, but end-to-end trained networks often result in hard to understand black-box classifiers or predictors. In this work, we present a novel method to identify whether a specific feature is relevant to a classifier’s decision or not. This relevance is determined at the level of the learned mapping, instead of for a single example. The approach does neither need retraining of the network nor information on intermediate results or gradients. The key idea of our approach builds upon concepts from causal inference. We interpret machine learning in a structural causal model and use Reichenbach’s common cause principle to infer whether a feature is relevant. We demonstrate empirically that the method is able to successfully evaluate the relevance of given features on three real-life data sets, namely MS COCO, CUB200 and HAM10000.

Kelvin Wong, Qiang Zhang, Ming Liang, Bin Yang, Renjie Liao, Abbas Sadat, Raquel Urtasun

We present a novel method for testing the safety of self-driving vehicles in simulation. We propose an alternative to sensor simulation, as sensor simulation is expensive and has large domain gaps. Instead, we directly simulate the outputs of the self-driving vehicle’s perception and prediction system, enabling realistic motion planning testing. Specifically, we use paired data in the form of ground truth labels and real perception and prediction outputs to train a model that predicts what the online system will produce. Importantly, the inputs to our system consists of high definition maps, bounding boxes, and trajectories, which can be easily sketched by a test engineer in a matter of minutes. This makes our approach a much more scalable solution. Quantitative results on two large-scale datasets demonstrate that we can realistically test motion planning using our simulations

Xiao Zhang, Rui Zhao, Yu Qiao, Hongsheng Li

Deep neural networks have achieved remarkable successes in learning feature representations for visual classification. However, deep features learned by the softmax cross-entropy loss generally show excessive intra-class variations. We argue that, because the traditional softmax losses aim to optimize only the relative differences between intra-class and inter-class distances (logits), it cannot obtain representative class prototypes (class weights/centers) to regularize intra-class distances, even when the training is converged. Previous efforts mitigate this problem by introducing auxiliary regularization losses. But these modified losses mainly focus on optimizing intra-class compactness, while ignoring keeping reasonable relations between different class prototypes. These lead to weak models and eventually limit their performance. To address this problem, this paper introduces a novel Radial Basis Function (RBF) distances to replace the commonly used inner products in the softmax loss function, such that it can adaptively assign losses to regularize the intra-class and inter-class distances by reshaping the relative differences, and thus creating more representative prototypes of classes to improve optimization. The proposed RBF-Softmax loss function not only effectively reduces intra-class distances, stabilizes the training behavior, and reserves ideal relations between prototypes, but also significantly improves the testing performance. Experiments on visual recognition benchmarks including MNIST, CIFAR-10/100, and ImageNet demonstrate that the proposed RBF-Softmax achieves better results than cross-entropy and other state-of-the-art classification losses. The code is at https://github.com/2han9x1a0release/RBF-Softmax.

Yuan Xue, Zihan Zhou, Xiaolei Huang

In architecture and computer-aided design, wireframes (i.e., line-based models) are widely used as basic 3D models for design evaluation and fast design iterations. However, unlike a full design file, a wireframe model lacks critical information, such as detailed shape, texture, and materials, needed by a conventional renderer to produce 2D renderings of the objects or scenes. In this paper, we bridge the information gap by generating photo-realistic rendering of indoor scenes from wireframe models in an image translation framework. While existing image synthesis methods can generate visually pleasing images for common objects such as faces and birds, these methods do not explicitly model and preserve essential structural constraints in a wireframe model, such as junctions, parallel lines, and planar surfaces. To this end, we propose a novel model based on a structure-appearance joint representation learned from both images and wireframes. In our model, structural constraints are explicitly enforced by learning a joint representation in a shared encoder network that must support the generation of both images and wireframes. Experiments on a wireframe-scene dataset show that our wireframe-to-image translation model significantly outperforms the state-of-the-art methods in both visual quality and structural integrity of generated images.

Senthil Purushwalkam, Tian Ye, Saurabh Gupta, Abhinav Gupta

In this paper, we focus on the task of extracting visual correspondences across videos. Given a query video clip from an action class, we aim to align it with training videos in space and time. Obtaining training data for such a fine-grained alignment task is challenging and often ambiguous. Hence, we propose a novel alignment procedure that learns such correspondence in space and time via cross video cycle-consistency. During training, given a pair of videos, we compute cycles that connect patches in a given frame in the first video by matching through frames in the second video. Cycles that connect overlapping patches together are encouraged to score higher than cycles that connect non-overlapping patches. Our experiments on the Penn Action and Pouring datasets demonstrate that the proposed method can successfully learn to correspond semantically similar patches across videos, and learns representations that are sensitive to object and action states.

Sunnie S. Y. Kim, Nicholas Kolkin, Jason Salavon, Gregory Shakhnarovich

Both geometry and texture are fundamental aspects of visual style. Existing style transfer methods, however, primarily focus on texture, almost entirely ignoring geometry. We propose deformable style transfer (DST), an optimization-based approach that jointly stylizes the texture and geometry of a content image to better match a style image. Unlike previous geometry-aware stylization methods, our approach is neither restricted to a particular domain (such as human faces), nor does it require training sets of matching style/content pairs. We demonstrate our method on a diverse set of content and style images including portraits, animals, objects, scenes, and paintings. Code has been made publicly available at https://github.com/sunniesuhyoung/DST.

Maximilian Augustin, Alexander Meinke, Matthias Hein

Neural networks have led to major improvements in image classification but suffer from being non-robust to adversarial changes, unreliable uncertainty estimates on out-distribution samples and their inscrutable black-box decisions. In this work we propose RATIO, a training procedure for Robustness via Adversarial Training on In- and Out-distribution, which leads to robust models with reliable and robust confidence estimates on the out-distribution. RATIO has similar generative properties to adversarial training so that visual counterfactuals produce class specific features. While adversarial training comes at the price of lower clean accuracy, RATIO achieves state-of-the-art $l_2$-adversarial robustness on CIFAR10 and maintains better clean accuracy.

Roei Herzig, Amir Bar, Huijuan Xu, Gal Chechik, Trevor Darrell, Amir Globerson

Generating realistic images of complex visual scenes becomes challenging when one wishes to control the structure of the generated images. Previous approaches showed that scenes with few entities can be controlled using scene graphs, but this approach struggles as the complexity of the graph (the number of objects and edges) increases. In this work, we show that one limitation of current methods is their inability to capture semantic equivalence in graphs. We present a novel model that addresses these issues by learning canonical graph representations from the data, resulting in improved image generation for complex visual scenes. Our model demonstrates improved empirical performance on large scene graphs, robustness to noise in the input scene graph, and generalization on semantically equivalent graphs. Finally, we show improved performance of the model on three different benchmarks: Visual Genome, COCO, and CLEVR.

Ke Han, Yan Huang, Zerui Chen, Liang Wang, Tieniu Tan

Low-resolution person re-identification (LR re-id) is a challenging task with low-resolution probes and high-resolution gallery images. To address the resolution mismatch, existing methods typically recover missing details for low-resolution probes by super-resolution. However, they usually pre-specify fixed scale factors for all images, and ignore the fact that choosing a preferable scale factor for certain image content probably greatly benefits the identification. In this paper, we propose a novel Prediction, Recovery and Identification (PRI) model for LR re-id, which adaptively recovers missing details by predicting a preferable scale factor based on the image content. To deal with the lack of ground-truth optimal scale factors, our model contains a self-supervised scale factor metric that automatically generates dynamic soft labels. The generated labels indicate probabilities that each scale factor is optimal, which are used as guidance to enhance the content-aware scale factor prediction. Consequently, our model can more accurately predict and recover the content-aware details, and achieve state-of-the-art performances on four LR re-id datasets.