CVPR 2018 Papers

Weakly-supervised semantic segmentation under image tags supervision is a challenging task as it directly associates high-level semantic to low-level appearance. To bridge this gap, in this paper, we propose an iterative bottom-up and top-down framework which alternatively expands object regions and optimizes segmentation network. We start from initial localization produced by classification networks. While classification networks are only responsive to small and coarse discriminative object regions, we argue that, these regions contain significant common features about objects. So in the bottom-up step, we mine common object features from the initial localization and expand object regions with the mined features. To supplement non-discriminative regions, saliency maps are then considered under Bayesian framework to refine the object regions. Then in the top-down step, the refined object regions are used as supervision to train the segmentation network and to predict object masks. These object masks provide more accurate localization and contain more regions of object. Further, we take these object masks as initial localization and mine common object features from them. These processes are conducted iteratively to progressively produce fine object masks and optimize segmentation networks. Experimental results on Pascal VOC 2012 dataset demonstrate that the proposed method outperforms previous state-of-the-art methods by a large margin.

This paper studies the problem of learning image semantic segmentation networks only using image-level labels as supervision, which is important since it can significantly reduce human annotation efforts. Recent state-of-the-art methods on this problem first infer the sparse and discriminative regions for each object class using a deep classification network, then train semantic a segmentation network using the discriminative regions as supervision. Inspired by the traditional image segmentation methods of seeded region growing, we propose to train a semantic segmentation network starting from the discriminative regions and progressively increase the pixel-level supervision using by seeded region growing. The seeded region growing module is integrated in a deep segmentation network and can benefit from deep features. Different from conventional deep networks which have fixed/static labels, the proposed weakly-supervised network generates new labels using the contextual information within an image. The proposed method significantly outperforms the weakly-supervised semantic segmentation methods using static labels, and obtains the state-of-the-art performance, which are 63.2% mIoU score on the PASCAL VOC 2012 test set and 26.0% mIoU score on the COCO dataset.

Facial action unit (AU) intensity estimation plays an important role in affective computing and human-computer interaction. Recent works have introduced deep neural networks for AU intensity estimation, but they require a large amount of intensity annotations. AU annotation needs strong domain expertise and it is expensive to construct a large database to learn deep models. We propose a novel knowledge-based semi-supervised deep convolutional neural network for AU intensity estimation with extremely limited AU annotations. Only the intensity annotations of peak and valley frames in training sequences are needed. To provide additional supervision for model learning, we exploit naturally existing constraints on AUs, including relative appearance similarity, temporal intensity ordering, facial symmetry, and contrastive appearance difference. Experimental evaluations are performed on two public benchmark databases. With around 2% of intensity annotations in FERA 2015 and around 1% in DISFA for training, our method can achieve comparable or even better performance than the state-of-the-art methods which use 100% of intensity annotations in the training set.

In this work, we address the task of weakly-supervised human action segmentation in long, untrimmed videos. Recent methods have relied on expensive learning models, such as Recurrent Neural Networks (RNN) and Hidden Markov Models (HMM). However, these methods suffer from expensive computational cost, thus are unable to be deployed in large scale. To overcome the limitations, the keys to our design are efficiency and scalability. We propose a novel action modeling framework, which consists of a new temporal convolutional network, named Temporal Convolutional Feature Pyramid Network (TCFPN), for predicting frame-wise action labels, and a novel training strategy for weakly-supervised sequence modeling, named Iterative Soft Boundary Assignment (ISBA), to align action sequences and update the network in an iterative fashion. The proposed framework is evaluated on two benchmark datasets, Breakfast and Hollywood Extended, with four different evaluation metrics. Extensive experimental results show that our methods achieve competitive or superior performance to state-of-the-art methods.

Human body part parsing, or human semantic part segmentation, is fundamental to many computer vision tasks. In conventional semantic segmentation methods, the ground truth segmentations are provided, and fully convolutional networks (FCN) are trained in an end-to-end scheme. Although these methods have demonstrated impressive results, their performance highly depends on the quantity and quality of training data. In this paper, we present a novel method to generate synthetic human part segmentation data using easily-obtained human keypoint annotations. Our key idea is to exploit the anatomical similarity among human to transfer the parsing results of a person to another person with similar pose. Using these estimated results as additional training data, our semi-supervised model outperforms its strong-supervised counterpart by 6 mIOU on the PASCAL-Person-Part dataset, and we achieve state-of-the-art human parsing results. Our approach is general and can be readily extended to other object/animal parsing task assuming that their anatomical similarity can be annotated by keypoints. The proposed model and accompanying source code will be made publicly available.

In this paper, we propose a novel weakly supervised model, Multi-scale Anchored Transformer Network (MATN), to accurately localize free-form textual phrases with only image-level supervision. The proposed MATN takes region proposals as localization anchors, and learns a multi-scale correspondence network to continuously search for phrase regions referring to the anchors. In this way, MATN can exploit useful cues from these anchors to reliably reason about locations of the regions described by the phrases given only image-level supervision. Through differentiable sampling on image spatial feature maps, MATN introduces a novel training objective to simultaneously minimize a contrastive reconstruction loss between different phrases from a single image and a set of triplet losses among multiple images with similar phrases. Superior to existing region proposal based methods, MATN searches for the optimal bounding box over the entire feature map instead of selecting a sub-optimal one from discrete region proposals. We evaluate MATN on the Flickr30K Entities and ReferItGame datasets. The experimental results show that MATN significantly outperforms the state-of-the-art methods.

Many new applications in drug discovery and functional genomics require capturing the morphology of individual imaged cells as comprehensively as possible rather than measuring one particular feature. In these so-called profiling experiments, the goal is to compare populations of cells treated with different chemicals or genetic perturbations in order to identify biomedically important similarities. Deep convolutional neural networks (CNNs) often make excellent feature extractors but require ground truth for training; this is rarely available in biomedical profiling experiments. We therefore propose to train CNNs based on a weakly supervised approach, where the network aims to classify each treatment against all others. Using this network as a feature extractor performed comparably to a network trained on non-biological, natural images on a chemical screen benchmark task, and improved results significantly on a more challenging genetic benchmark presented for the first time.

Weakly supervised instance segmentation with image-level labels, instead of expensive pixel-level masks, remains unexplored. In this paper, we tackle this challenging problem by exploiting class peak responses to enable a classification network for instance mask extraction. With image labels supervision only, CNN classifiers in a fully convolutional manner can produce class response maps, which specify classification confidence at each image location. We observed that local maximums, i.e., peaks, in a class response map typically correspond to strong visual cues residing inside each instance. Motivated by this, we first design a process to stimulate peaks to emerge from a class response map. The emerged peaks are then back-propagated and effectively mapped to highly informative regions of each object instance, such as instance boundaries. We refer to the above maps generated from class peak responses as Peak Response Maps (PRMs). PRMs provide a fine-detailed instance-level representation, which allows instance masks to be extracted even with some off-the-shelf methods. To the best of our knowledge, we for the first time report results for the challenging image-level supervised instance segmentation task. Extensive experiments show that our method also boosts weakly supervised pointwise localization as well as semantic segmentation performance, and reports state-of-the-art results on popular benchmarks, including PASCAL VOC 2012 and MS COCO.

Current works on facial action unit (AU) recognition typically require fully AU-annotated facial images for supervised AU classifier training. AU annotation is a time-consuming, expensive, and error-prone process. While AUs are hard to annotate, facial expression is relatively easy to label. Furthermore, there exist strong probabilistic dependencies between expressions and AUs as well as dependencies among AUs. Such dependencies are referred to as domain knowledge. In this paper, we propose a novel AU recognition method that learns AU classifiers from domain knowledge and expression-annotated facial images through adversarial training. Specifically, we first generate pseudo AU labels according to the probabilistic dependencies between expressions and AUs as well as correlations among AUs summarized from domain knowledge. Then we propose a weakly supervised AU recognition method via an adversarial process, in which we simultaneously train two models: a recognition model R, which learns AU classifiers, and a discrimination model D, which estimates the probability that AU labels generated from domain knowledge rather than the recognized AU labels from R. The training procedure for R maximizes the probability of D making a mistake. By leveraging the adversarial mechanism, the distribution of recognized AUs is closed to AU prior distribution from domain knowledge. Furthermore, the proposed weakly supervised AU recognition can be extended to semi-supervised learning scenarios with partially AU-annotated images. Experimental results on three benchmark databases demonstrate that the proposed method successfully leverages the summarized domain knowledge to weakly supervised AU classifier learning through an adversarial process, and thus achieves state-of-the-art performance.

Automatic assessment of sentiment from visual content has gained considerable attention with the increasing tendency of expressing opinions on-line. In this paper, we solve the problem of visual sentiment analysis using the high-level abstraction in the recognition process. Existing methods based on convolutional neural networks learn sentiment representations from the holistic image appearance. However, different image regions can have a different influence on the intended expression. This paper presents a weakly supervised coupled convolutional network with two branches to leverage the localized information. The first branch detects a sentiment specific soft map by training a fully convolutional network with the cross spatial pooling strategy, which only requires image-level labels, thereby significantly reducing the annotation burden. The second branch utilizes both the holistic and localized information by coupling the sentiment map with deep features for robust classification. We integrate the sentiment detection and classification branches into a unified deep framework and optimize the network in an end-to-end manner. Extensive experiments on six benchmark datasets demonstrate that the proposed method performs favorably against the state-ofthe-art methods for visual sentiment analysis.

We propose a weakly supervised temporal action localization algorithm on untrimmed videos using convolutional neural networks. Our algorithm learns from video-level class labels and predicts temporal intervals of human actions with no requirement of temporal localization annotations. We design our network to identify a sparse subset of key segments associated with target actions in a video using an attention module and fuse the key segments through adaptive temporal pooling. Our loss function is comprised of two terms that minimize the video-level action classification error and enforce the sparsity of the segment selection. At inference time, we extract and score temporal proposals using temporal class activations and class-agnostic attentions to estimate the time intervals that correspond to target actions. The proposed algorithm attains state-of-the-art results on the THUMOS14 dataset and outstanding performance on ActivityNet1.3 even with its weak supervision.

We present Wasserstein introspective neural networks (WINN) that are both a generator and a discriminator within a single model. WINN provides a significant improvement over the recent introspective neural networks (INN) method by enhancing INN's generative modeling capability. WINN has three interesting properties: (1) A mathematical connection between the formulation of the INN algorithm and that of Wasserstein generative adversarial networks (WGAN) is made. (2) The explicit adoption of the Wasserstein distance into INN results in a large enhancement to INN, achieving compelling results even with a single classifier --- e.g., providing nearly a 20 times reduction in model size over INN for unsupervised generative modeling. (3) When applied to supervised classification, WINN also gives rise to improved robustness against adversarial examples in terms of the error reduction. In the experiments, we report encouraging results on unsupervised learning problems including texture, face, and object modeling, as well as a supervised classification task against adversarial attacks.

People detection methods are highly sensitive to occlusions between pedestrians, which are extremely frequent in many situations where cameras have to be mounted at a limited height. The reduction of camera prices allows for the generalization of static multi-camera set-ups. Using joint visual information from multiple synchronized cameras gives the opportunity to improve detection performance. In this paper, we present a new large-scale and high-resolution dataset. It has been captured with seven static cameras in a public open area, and unscripted dense groups of pedestrians standing and walking. Together with the camera frames, we provide an accurate joint (extrinsic and intrinsic) calibration, as well as 7 series of 400 annotated frames for detection at a rate of 2 frames per second. This results in over 40,000 bounding boxes delimiting every person present in the area of interest, for a total of more than 300 individuals. We provide a series of benchmark results using baseline algorithms published over the recent months for multi-view detection with deep neural networks, and trajectory estimation using a non-Markovian model.

Weakly-supervised object detection has attracted much attention lately, since it does not require bounding box annotations for training. Although significant progress has also been made, there is still a large gap in performance between weakly-supervised and fully-supervised object detection. Recently, some works use pseudo ground-truths which are generated by a weakly-supervised detector to train a supervised detector. Such approaches incline to find the most representative parts of objects, and only seek one ground-truth box per class even though many same-class instances exist. To overcome these issues, we propose a weakly-supervised to fully-supervised framework, where a weakly-supervised detector is implemented using multiple instance learning. Then, we propose a pseudo ground-truth excavation (PGE) algorithm to find the pseudo ground-truth of each instance in the image. Moreover, the pseudo ground-truth adaptation (PGA) algorithm is designed to further refine the pseudo ground-truths from PGE. Finally, we use these pseudo ground-truths to train a fully-supervised detector. Extensive experiments on the challenging PASCAL VOC 2007 and 2012 benchmarks strongly demonstrate the effectiveness of our framework. We obtain 52.4% and 47.8% mAP on VOC2007 and VOC2012 respectively, a significant improvement over previous state-of-the-art methods.

Accurate detection of objects in 3D point clouds is a central problem in many applications, such as autonomous navigation, housekeeping robots, and augmented/virtual reality. To interface a highly sparse LiDAR point cloud with a region proposal network (RPN), most existing efforts have focused on hand-crafted feature representations, for example, a bird's eye view projection. In this work, we remove the need of manual feature engineering for 3D point clouds and propose VoxelNet, a generic 3D detection network that unifies feature extraction and bounding box prediction into a single stage, end-to-end trainable deep network. Specifically, VoxelNet divides a point cloud into equally spaced 3D voxels and transforms a group of points within each voxel into a unified feature representation through the newly introduced voxel feature encoding (VFE) layer. In this way, the point cloud is encoded as a descriptive volumetric representation, which is then connected to a RPN to generate detections. Experiments on the KITTI car detection benchmark show that VoxelNet outperforms the state-of-the-art LiDAR based 3D detection methods by a large margin. Furthermore, our network learns an effective discriminative representation of objects with various geometries, leading to encouraging results in 3D detection of pedestrians and cyclists, based on only LiDAR.

The study of algorithms to automatically answer visual questions currently is motivated by visual question answering (VQA) datasets constructed in artificial VQA settings. We propose VizWiz, the first goal-oriented VQA dataset arising from a natural VQA setting. VizWiz consists of 31,000 visual questions originating from blind people who each took a picture using a mobile phone and recorded a spoken question about it, together with 10 crowdsourced answers per visual question. VizWiz differs from the many existing VQA datasets because (1) images are captured by blind photographers and so are often poor quality, (2) questions are spoken and so are more conversational, and (3) often visual questions cannot be answered. Evaluation of modern algorithms for answering visual questions and deciding if a visual question is answerable reveals that VizWiz is a challenging dataset. We introduce this dataset to encourage a larger community to develop more generalized algorithms that can assist blind people.

As two of the five traditional human senses (sight, hearing, taste, smell, and touch), vision and sound are basic sources through which humans understand the world. Often correlated during natural events, these two modalities combine to jointly affect human perception. In this paper, we pose the task of generating sound given visual input. Such capabilities could help enable applications in virtual reality (generating sound for virtual scenes automatically) or provide additional accessibility to images or videos for people with visual impairments. As a first step in this direction, we apply learning-based methods to generate raw waveform samples given input video frames. We evaluate our models on a dataset of videos containing a variety of sounds (such as ambient sounds and sounds from people/animals). Our experiments show that the generated sounds are fairly realistic and have good temporal synchronization with the visual inputs.

The collaborative reasoning for understanding each image-question pair is very critical but under-explored for an interpretable Visual Question Answering (VQA) system. Although very recent works also tried the explicit compositional processes to assemble multiple sub-tasks embedded in the questions, their models heavily rely on the annotations or hand-crafted rules to obtain valid reasoning layout, leading to either heavy labor or poor performance on composition reasoning. In this paper, to enable global context reasoning for better aligning image and language domains in diverse and unrestricted cases, we propose a novel reasoning network called Adversarial Composition Modular Network (ACMN). This network comprises of two collaborative modules: i) an adversarial attention module to exploit the local visual evidence for each word parsed from the question; ii) a residual composition module to compose the previously mined evidence. Given a dependency parse tree for each question, the adversarial attention module progressively discovers salient regions of one word by densely combining regions of child word nodes in an adversarial manner. Then residual composition module merges the hidden representations of an arbitrary number of children through sum pooling and residual connection. Our ACMN is thus capable of building an interpretable VQA system that gradually dives the image cues following a question-driven reasoning route and makes global reasoning by incorporating the learned knowledge of all attention modules in a principled manner. Experiments on relational datasets demonstrate the superiority of our ACMN and visualization results show the explainable capability of our reasoning system.

Visual question answering (VQA) and visual question generation (VQG) are two trending topics in the computer vision, but they are usually explored separately despite their intrinsic complementary relationship. In this paper, we propose an end-to-end unified model, the Invertible Question Answering Network (iQAN), to introduce question generation as a dual task of question answering to improve the VQA performance. With our proposed invertible bilinear fusion module and parameter sharing scheme, our iQAN can accomplish VQA and its dual task VQG simultaneously. By jointly trained on two tasks with our proposed dual regularizers~(termed as Dual Training), our model has a better understanding of the interactions among images, questions and answers. After training, iQAN can take either question or answer as input, and output the counterpart. Evaluated on the CLEVR and VQA2 datasets, our iQAN improves the top-1 accuracy of the prior art MUTAN VQA method by 1.33% and 0.88% (absolute increase). We also show that our proposed dual training framework can consistently improve model performances of many popular VQA architectures.