The ability for computational agents to reason about the high-level content of real world scene images is important for many applications. Existing attempts at addressing the problem of complex scene understanding lack representational power, efficiency, and the ability to create robust meta-knowledge about scenes. In this paper, we introduce scenarios as a new way of representing scenes. The scenario is a simple, low-dimensional, data-driven representation consisting of sets of frequently co-occurring objects and is useful for a wide range of scene understanding tasks. We learn scenarios from data using a novel matrix factorization method which we integrate into a new neural network architecture, the ScenarioNet. Using ScenarioNet, we can recover semantic information about real world scene images at three levels of granularity: 1) scene categories, 2) scenarios, and 3) objects. Training a single ScenarioNet model enables us to perform scene classification, scenario recognition, multi-object recognition, content-based scene image retrieval, and content-based image comparison. In addition to solving many tasks in a single, unified framework, ScenarioNet is more computationally efficient than other CNNs because it requires significantly fewer parameters while achieving similar performance on benchmark tasks and is more interpretable because it produces explanations when making decisions. We validate the utility of scenarios and ScenarioNet on a diverse set of scene understanding tasks on several benchmark datasets.

Click to Read Paper
Humans are able to identify a referred visual object in a complex scene via a few rounds of natural language communications. Success communication requires both parties to engage and learn to adapt for each other. In this paper, we introduce an interactive training method to improve the natural language conversation system for a visual grounding task. During interactive training, both agents are reinforced by the guidance from a common reward function. The parametrized reward function also cooperatively updates itself via interactions, and contribute to accomplishing the task. We evaluate the method on GuessWhat?! visual grounding task, and significantly improve the task success rate. However, we observe language drifting problem during training and propose to use reward engineering to improve the interpretability for the generated conversations. Our result also indicates evaluating goal-ended visual conversation tasks require semantic relevant metrics beyond task success rate.

* NIPS 2017 - Visually-Grounded Interaction and Language (ViGIL) Workshop
Click to Read Paper
In this paper, we propose the Self-Attention Generative Adversarial Network (SAGAN) which allows attention-driven, long-range dependency modeling for image generation tasks. Traditional convolutional GANs generate high-resolution details as a function of only spatially local points in lower-resolution feature maps. In SAGAN, details can be generated using cues from all feature locations. Moreover, the discriminator can check that highly detailed features in distant portions of the image are consistent with each other. Furthermore, recent work has shown that generator conditioning affects GAN performance. Leveraging this insight, we apply spectral normalization to the GAN generator and find that this improves training dynamics. The proposed SAGAN achieves the state-of-the-art results, boosting the best published Inception score from 36.8 to 52.52 and reducing Frechet Inception distance from 27.62 to 18.65 on the challenging ImageNet dataset. Visualization of the attention layers shows that the generator leverages neighborhoods that correspond to object shapes rather than local regions of fixed shape.

Click to Read Paper
We present Optimal Transport GAN (OT-GAN), a variant of generative adversarial nets minimizing a new metric measuring the distance between the generator distribution and the data distribution. This metric, which we call mini-batch energy distance, combines optimal transport in primal form with an energy distance defined in an adversarially learned feature space, resulting in a highly discriminative distance function with unbiased mini-batch gradients. Experimentally we show OT-GAN to be highly stable when trained with large mini-batches, and we present state-of-the-art results on several popular benchmark problems for image generation.

Click to Read Paper
Multispectral pedestrian detection is essential for around-the-clock applications, e.g., surveillance and autonomous driving. We deeply analyze Faster R-CNN for multispectral pedestrian detection task and then model it into a convolutional network (ConvNet) fusion problem. Further, we discover that ConvNet-based pedestrian detectors trained by color or thermal images separately provide complementary information in discriminating human instances. Thus there is a large potential to improve pedestrian detection by using color and thermal images in DNNs simultaneously. We carefully design four ConvNet fusion architectures that integrate two-branch ConvNets on different DNNs stages, all of which yield better performance compared with the baseline detector. Our experimental results on KAIST pedestrian benchmark show that the Halfway Fusion model that performs fusion on the middle-level convolutional features outperforms the baseline method by 11% and yields a missing rate 3.5% lower than the other proposed architectures.

* 13 pages, 8 figures, BMVC 2016 oral
Click to Read Paper
Tracking Facial Points in unconstrained videos is challenging due to the non-rigid deformation that changes over time. In this paper, we propose to exploit incremental learning for person-specific alignment in wild conditions. Our approach takes advantage of part-based representation and cascade regression for robust and efficient alignment on each frame. Unlike existing methods that usually rely on models trained offline, we incrementally update the representation subspace and the cascade of regressors in a unified framework to achieve personalized modeling on the fly. To alleviate the drifting issue, the fitting results are evaluated using a deep neural network, where well-aligned faces are picked out to incrementally update the representation and fitting models. Both image and video datasets are employed to valid the proposed method. The results demonstrate the superior performance of our approach compared with existing approaches in terms of fitting accuracy and efficiency.

* British Machine Vision Conference (BMVC), 2016
Click to Read Paper
In this paper, we propose a novel visual tracking framework that intelligently discovers reliable patterns from a wide range of video to resist drift error for long-term tracking tasks. First, we design a Discrete Fourier Transform (DFT) based tracker which is able to exploit a large number of tracked samples while still ensures real-time performance. Second, we propose a clustering method with temporal constraints to explore and memorize consistent patterns from previous frames, named as reliable memories. By virtue of this method, our tracker can utilize uncontaminated information to alleviate drifting issues. Experimental results show that our tracker performs favorably against other state of-the-art methods on benchmark datasets. Furthermore, it is significantly competent in handling drifts and able to robustly track challenging long videos over 4000 frames, while most of others lose track at early frames.

Click to Read Paper
We consider the problem of image-to-video translation, where an input image is translated into an output video containing motions of a single object. Recent methods for such problems typically train transformation networks to generate future frames conditioned on the structure sequence. Parallel work has shown that short high-quality motions can be generated by spatiotemporal generative networks that leverage temporal knowledge from the training data. We combine the benefits of both approaches and propose a two-stage generation framework where videos are generated from structures and then refined by temporal signals. To model motions more efficiently, we train networks to learn residual motion between the current and future frames, which avoids learning motion-irrelevant details. We conduct extensive experiments on two image-to-video translation tasks: facial expression retargeting and human pose forecasting. Superior results over the state-of-the-art methods on both tasks demonstrate the effectiveness of our approach.

* 17 pages, 8 figures, 4 tables, accepted by ECCV 2018
Click to Read Paper
Random data augmentation is a critical technique to avoid overfitting in training deep neural network models. However, data augmentation and network training are usually treated as two isolated processes, limiting the effectiveness of network training. Why not jointly optimize the two? We propose adversarial data augmentation to address this limitation. The main idea is to design an augmentation network (generator) that competes against a target network (discriminator) by generating `hard' augmentation operations online. The augmentation network explores the weaknesses of the target network, while the latter learns from `hard' augmentations to achieve better performance. We also design a reward/penalty strategy for effective joint training. We demonstrate our approach on the problem of human pose estimation and carry out a comprehensive experimental analysis, showing that our method can significantly improve state-of-the-art models without additional data efforts.

* CVPR 2018
Click to Read Paper
We propose a novel method for real-time face alignment in videos based on a recurrent encoder-decoder network model. Our proposed model predicts 2D facial point heat maps regularized by both detection and regression loss, while uniquely exploiting recurrent learning at both spatial and temporal dimensions. At the spatial level, we add a feedback loop connection between the combined output response map and the input, in order to enable iterative coarse-to-fine face alignment using a single network model, instead of relying on traditional cascaded model ensembles. At the temporal level, we first decouple the features in the bottleneck of the network into temporal-variant factors, such as pose and expression, and temporal-invariant factors, such as identity information. Temporal recurrent learning is then applied to the decoupled temporal-variant features. We show that such feature disentangling yields better generalization and significantly more accurate results at test time. We perform a comprehensive experimental analysis, showing the importance of each component of our proposed model, as well as superior results over the state of the art and several variations of our method in standard datasets.

* International Journal of Computer Vision. arXiv admin note: text overlap with arXiv:1608.05477
Click to Read Paper
We propose a novel recurrent encoder-decoder network model for real-time video-based face alignment. Our proposed model predicts 2D facial point maps regularized by a regression loss, while uniquely exploiting recurrent learning at both spatial and temporal dimensions. At the spatial level, we add a feedback loop connection between the combined output response map and the input, in order to enable iterative coarse-to-fine face alignment using a single network model. At the temporal level, we first decouple the features in the bottleneck of the network into temporal-variant factors, such as pose and expression, and temporal-invariant factors, such as identity information. Temporal recurrent learning is then applied to the decoupled temporal-variant features, yielding better generalization and significantly more accurate results at test time. We perform a comprehensive experimental analysis, showing the importance of each component of our proposed model, as well as superior results over the state-of-the-art in standard datasets.

* European Conference on Computer Vision (ECCV), 2016
Click to Read Paper
We address the problem of reconstructing and analyzing surveillance videos using compressive sensing. We develop a new method that performs video reconstruction by low rank and sparse decomposition adaptively. Background subtraction becomes part of the reconstruction. In our method, a background model is used in which the background is learned adaptively as the compressive measurements are processed. The adaptive method has low latency, and is more robust than previous methods. We will present experimental results to demonstrate the advantages of the proposed method.

* IEEE International Conference on Image Processing, ICIP 2013, Paper #1870
* Accepted ICIP 2013
Click to Read Paper
We design a new connectivity pattern for the U-Net architecture. Given several stacked U-Nets, we couple each U-Net pair through the connections of their semantic blocks, resulting in the coupled U-Nets (CU-Net). The coupling connections could make the information flow more efficiently across U-Nets. The feature reuse across U-Nets makes each U-Net very parameter efficient. We evaluate the coupled U-Nets on two benchmark datasets of human pose estimation. Both the accuracy and model parameter number are compared. The CU-Net obtains comparable accuracy as state-of-the-art methods. However, it only has at least 60% fewer parameters than other approaches.

* BMVC 2018 (Oral)
Click to Read Paper
Generating multi-view images from a single-view input is an essential yet challenging problem. It has broad applications in vision, graphics, and robotics. Our study indicates that the widely-used generative adversarial network (GAN) may learn "incomplete" representations due to the single-pathway framework: an encoder-decoder network followed by a discriminator network. We propose CR-GAN to address this problem. In addition to the single reconstruction path, we introduce a generation sideway to maintain the completeness of the learned embedding space. The two learning pathways collaborate and compete in a parameter-sharing manner, yielding considerably improved generalization ability to "unseen" dataset. More importantly, the two-pathway framework makes it possible to combine both labeled and unlabeled data for self-supervised learning, which further enriches the embedding space for realistic generations. The experimental results prove that CR-GAN significantly outperforms state-of-the-art methods, especially when generating from "unseen" inputs in wild conditions.

* 7 pages, 9 figures, accepted by IJCAI 2018
Click to Read Paper
Lifting is a common manual material handling task performed in the workplaces. It is considered as one of the main risk factors for Work-related Musculoskeletal Disorders. To improve work place safety, it is necessary to assess musculoskeletal and biomechanical risk exposures associated with these tasks, which requires very accurate 3D pose. Existing approaches mainly utilize marker-based sensors to collect 3D information. However, these methods are usually expensive to setup, time-consuming in process, and sensitive to the surrounding environment. In this study, we propose a multi-view based deep perceptron approach to address aforementioned limitations. Our approach consists of two modules: a "view-specific perceptron" network extracts rich information independently from the image of view, which includes both 2D shape and hierarchical texture information; while a "multi-view integration" network synthesizes information from all available views to predict accurate 3D pose. To fully evaluate our approach, we carried out comprehensive experiments to compare different variants of our design. The results prove that our approach achieves comparable performance with former marker-based methods, i.e. an average error of $14.72 \pm 2.96$ mm on the lifting dataset. The results are also compared with state-of-the-art methods on HumanEva-I dataset, which demonstrates the superior performance of our approach.

* FG2018, accepted as a long paper
Click to Read Paper
Iterative Hard Thresholding (IHT) is a class of projected gradient descent methods for optimizing sparsity-constrained minimization models, with the best known efficiency and scalability in practice. As far as we know, the existing IHT-style methods are designed for sparse minimization in primal form. It remains open to explore duality theory and algorithms in such a non-convex and NP-hard problem setting. In this paper, we bridge this gap by establishing a duality theory for sparsity-constrained minimization with $\ell_2$-regularized loss function and proposing an IHT-style algorithm for dual maximization. Our sparse duality theory provides a set of sufficient and necessary conditions under which the original NP-hard/non-convex problem can be equivalently solved in a dual formulation. The proposed dual IHT algorithm is a super-gradient method for maximizing the non-smooth dual objective. An interesting finding is that the sparse recovery performance of dual IHT is invariant to the Restricted Isometry Property (RIP), which is required by virtually all the existing primal IHT algorithms without sparsity relaxation. Moreover, a stochastic variant of dual IHT is proposed for large-scale stochastic optimization. Numerical results demonstrate the superiority of dual IHT algorithms to the state-of-the-art primal IHT-style algorithms in model estimation accuracy and computational efficiency.

Click to Read Paper
We consider an MRI reconstruction problem with input of k-space data at a very low undersampled rate. This can practically benefit patient due to reduced time of MRI scan, but it is also challenging since quality of reconstruction may be compromised. Currently, deep learning based methods dominate MRI reconstruction over traditional approaches such as Compressed Sensing, but they rarely show satisfactory performance in the case of low undersampled k-space data. One explanation is that these methods treat channel-wise features equally, which results in degraded representation ability of the neural network. To solve this problem, we propose a new model called MRI Cascaded Channel-wise Attention Network (MICCAN), highlighted by three components: (i) a variant of U-net with Channel-wise Attention (UCA) module, (ii) a long skip connection and (iii) a combined loss. Our model is able to attend to salient information by filtering irrelevant features and also concentrate on high-frequency information by enforcing low-frequency information bypassed to the final output. We conduct both quantitative evaluation and qualitative analysis of our method on a cardiac dataset. The experiment shows that our method achieves very promising results in terms of three common metrics on the MRI reconstruction with low undersampled k-space data.

* 4 pages, 2 figures
Click to Read Paper
In this paper, we propose quantized densely connected U-Nets for efficient visual landmark localization. The idea is that features of the same semantic meanings are globally reused across the stacked U-Nets. This dense connectivity largely improves the information flow, yielding improved localization accuracy. However, a vanilla dense design would suffer from critical efficiency issue in both training and testing. To solve this problem, we first propose order-K dense connectivity to trim off long-distance shortcuts; then, we use a memory-efficient implementation to significantly boost the training efficiency and investigate an iterative refinement that may slice the model size in half. Finally, to reduce the memory consumption and high precision operations both in training and testing, we further quantize weights, inputs, and gradients of our localization network to low bit-width numbers. We validate our approach in two tasks: human pose estimation and face alignment. The results show that our approach achieves state-of-the-art localization accuracy, but using ~70% fewer parameters, ~98% less model size and saving ~75% training memory compared with other benchmark localizers. The code is available at https://github.com/zhiqiangdon/CU-Net.

* ECCV2018
Click to Read Paper
Although Generative Adversarial Networks (GANs) have shown remarkable success in various tasks, they still face challenges in generating high quality images. In this paper, we propose Stacked Generative Adversarial Networks (StackGAN) aiming at generating high-resolution photo-realistic images. First, we propose a two-stage generative adversarial network architecture, StackGAN-v1, for text-to-image synthesis. The Stage-I GAN sketches the primitive shape and colors of the object based on given text description, yielding low-resolution images. The Stage-II GAN takes Stage-I results and text descriptions as inputs, and generates high-resolution images with photo-realistic details. Second, an advanced multi-stage generative adversarial network architecture, StackGAN-v2, is proposed for both conditional and unconditional generative tasks. Our StackGAN-v2 consists of multiple generators and discriminators in a tree-like structure; images at multiple scales corresponding to the same scene are generated from different branches of the tree. StackGAN-v2 shows more stable training behavior than StackGAN-v1 by jointly approximating multiple distributions. Extensive experiments demonstrate that the proposed stacked generative adversarial networks significantly outperform other state-of-the-art methods in generating photo-realistic images.

* In IEEE Trans. on Pattern Analysis and Machine Intelligence (TPAMI), 2018. (16 pages, 15 figures.)
Click to Read Paper
Synthesizing high-quality images from text descriptions is a challenging problem in computer vision and has many practical applications. Samples generated by existing text-to-image approaches can roughly reflect the meaning of the given descriptions, but they fail to contain necessary details and vivid object parts. In this paper, we propose Stacked Generative Adversarial Networks (StackGAN) to generate 256x256 photo-realistic images conditioned on text descriptions. We decompose the hard problem into more manageable sub-problems through a sketch-refinement process. The Stage-I GAN sketches the primitive shape and colors of the object based on the given text description, yielding Stage-I low-resolution images. The Stage-II GAN takes Stage-I results and text descriptions as inputs, and generates high-resolution images with photo-realistic details. It is able to rectify defects in Stage-I results and add compelling details with the refinement process. To improve the diversity of the synthesized images and stabilize the training of the conditional-GAN, we introduce a novel Conditioning Augmentation technique that encourages smoothness in the latent conditioning manifold. Extensive experiments and comparisons with state-of-the-arts on benchmark datasets demonstrate that the proposed method achieves significant improvements on generating photo-realistic images conditioned on text descriptions.

* ICCV 2017 Oral Presentation
Click to Read Paper