ICCV 2015 Papers

The log-likelihood energy term in popular model-fitting segmentation methods, e.g. Zhu&Yuille, Chan-Vese, GrabCut, is presented as a generalized "probabilistic K-means" energy for color space clustering. This interpretation reveals some limitations, e.g. over-fitting. We propose an alternative approach to color clustering using kernel K-means energy with well-known properties such as non-linear separation and scalability to higher-dimensional feature spaces. Our bound formulation for kernel K-means allows to combine general pair-wise feature clustering methods with image grid regularization using graph cuts, similarly to standard color model fitting techniques for segmentation. Unlike histogram or GMM fitting, our approach is closely related to average association and normalized cut. But, in contrast to previous pairwise clustering algorithms, our approach can incorporate any standard geometric regularization in the image domain. We analyze extreme cases for kernel bandwidth (e.g. Gini bias) and demonstrate effectiveness of KNN-based adaptive bandwidth strategies. Our kernel K-means approach to segmentation benefits from higher-dimensional features where standard model fitting fails.

In this paper, we are interested in enhancing the expressivity and robustness of part-based models for object representation, in the common scenario where the training data are based on 2D images. To this end, we propose scene-domain active part models (SDAPM), which reconstruct and characterize the 3D geometric statistics between object's parts in 3D scene-domain by using 2D training data in the image-domain alone. And on top of this, we explicitly model and handle occlusions in SDAPM. Together with the developed learning and inference algorithms, such a model provides rich object descriptions, including 2D object and parts localization, 3D landmark shape and camera viewpoint, which offers an effective representation to various image understanding tasks, such as object and parts detection, 3D landmark shape and viewpoint estimation from images. Experiments on the above tasks show that SDAPM outperforms previous part-based models, and thus demonstrates the potential of the proposed technique.

This paper contributes a new high quality dataset for person re-identification, named "Market-1501". Generally, current datasets: 1) are limited in scale; 2) consist of hand-drawn bboxes, which are unavailable under realistic settings; 3) have only one ground truth and one query image for each identity (close environment). To tackle these problems, the proposed Market-1501 dataset is featured in three aspects. First, it contains over 32,000 annotated bboxes, plus a distractor set of over 500K images, making it the largest person re-id dataset to date. Second, images in Market-1501 dataset are produced using the Deformable Part Model (DPM) as pedestrian detector. Third, our dataset is collected in an open system, where each identity has multiple images under each camera. As a minor contribution, inspired by recent advances in large-scale image search, this paper proposes an unsupervised Bag-of-Words descriptor. We view person re-identification as a special task of image search. In experiment, we show that the proposed descriptor yields competitive accuracy on VIPeR, CUHK03, and Market-1501 datasets, and is scalable on the large-scale 500k dataset.

We propose a novel algorithm for the task of supervised discriminative distance learning by nonlinearly embedding vectors into a low dimensional Euclidean space. We work in the challenging setting where supervision is with constraints on similar and dissimilar pairs while training. The proposed method is derived by an approximate kernelization of a linear Mahalanobis-like distance metric learning algorithm and can also be seen as a kernel neural network. The number of model parameters and test time evaluation complexity of the proposed method are O(dD) where D is the dimensionality of the input features and d is the dimension of the projection space -- this is in contrast to the usual kernelization methods as, unlike them, the complexity does not scale linearly with the number of training examples. We propose a stochastic gradient based learning algorithm which makes the method scalable (w.r.t. the number of training examples), while being nonlinear. We train the method with up to half a million training pairs of 4096 dimensional CNN features. We give empirical comparisons with relevant baselines on seven challenging datasets for the task of low dimensional semantic category based image retrieval.

A simple yet effective object descriptor for visual tracking is proposed in this paper. We first decompose the bounding box of a target object into multiple patches, which are described by color and gradient histograms. Then, we concatenate the features of the spatially ordered patches to represent the object appearance. Moreover, to alleviate the impacts of background information possibly included in the bounding box, we determine patch weights using random walk with restart (RWR) simulations. The patch weights represent the importance of each patch in the description of foreground information, and are used to construct an object descriptor, called spatially ordered and weighted patch (SOWP) descriptor. We incorporate the proposed SOWP descriptor into the structured output tracking framework. Experimental results demonstrate that the proposed algorithm yields significantly better performance than the state-of-the-art trackers on a recent benchmark dataset, and also excels in another recent benchmark dataset.

Saliency in Context (SALICON) is an ongoing effort that aims at understanding and predicting visual attention. Conventional saliency models typically rely on low-level image statistics to predict human fixations. While these models perform significantly better than chance, there is still a large gap between model prediction and human behavior. This gap is largely due to the limited capability of models in predicting eye fixations with strong semantic content, the so-called semantic gap. This paper presents a focused study to narrow the semantic gap with an architecture based on Deep Neural Network (DNN). It leverages the representational power of high-level semantics encoded in DNNs pretrained for object recognition. Two key components are fine-tuning the DNNs fully convolutionally with an objective function based on the saliency evaluation metrics, and integrating information at different image scales. We compare our method with 14 saliency models on 6 public eye tracking benchmark datasets. Results demonstrate that our DNNs can automatically learn features particularly for saliency prediction that surpass by a big margin the state-of-the-art. In addition, our model ranks top to date under all seven metrics on the MIT300 challenge set.

Classical multi-image super-resolution (SR) algorithms, designed for CCD cameras, assume that the motion among the images is global. But CMOS sensors that have increasingly started to replace their more expensive CCD counterparts in many applications do not respect this assumption if there is a motion of the camera relative to the scene during the exposure duration of an image because of the row-wise acquisition mechanism. In this paper, we study the hitherto unexplored topic of multi-image SR in CMOS cameras. We initially develop an SR observation model that accounts for the row-wise distortions called the ``rolling shutter'' (RS) effect observed in images captured using non-stationary CMOS cameras. We then propose a unified RS-SR framework to obtain an RS-free high-resolution image (and the row-wise motion) from distorted low-resolution images. We demonstrate the efficacy of the proposed scheme using synthetic data as well as real images captured using a hand-held CMOS camera. Quantitative and qualitative assessments reveal that our method significantly advances the state-of-the-art.

Recently, it has been shown that excellent results can be achieved in both facial landmark localization and pose-invariant face recognition. These breakthroughs are attributed to the efforts of the community to manually annotate facial images in many different poses and to collect 3D facial data. In this paper, we propose a novel method for joint frontal view reconstruction and landmark localization using a small set of frontal images only. By observing that the frontal facial image is the one having the minimum rank of all different poses, an appropriate model which is able to jointly recover the frontalized version of the face as well as the facial landmarks is devised. To this end, a suitable optimization problem, involving the minimization of the nuclear norm and the matrix l1 norm is solved. The proposed method is assessed in frontal face reconstruction, face landmark localization, pose-invariant face recognition, and face verification in unconstrained conditions. The relevant experiments have been conducted on 8 databases. The experimental results demonstrate the effectiveness of the proposed method in comparison to the state-of-the-art methods for the target problems.

Lighting variation and uneven feature distribution are main challenges for indoor RGB-D visual odometry where color information is often combined with depth information. To meet the challenges, we fuse point and line features to form a robust odometry algorithm. Line features are abundant indoors and less sensitive to lighting change than points. We extract 3D points and lines from RGB-D data, analyze their measurement uncertainties, and compute camera motion using maximum likelihood estimation. We prove that fusing points and lines produces smaller motion estimate uncertainty than using either feature type alone. In experiments we compare our method with state-of-the-art methods including a keypoint-based approach and a dense visual odometry algorithm. Our method outperforms the counterparts under both constant and varying lighting conditions. Specifically, our method achieves an average translational error that is 34.9% smaller than the counterparts, when tested using public datasets.

Principal Component Analysis (PCA) is the most widely used tool for linear dimensionality reduction and clustering. Still it is highly sensitive to outliers and does not scale well with respect to the number of data samples. Robust PCA solves the first issue with a sparse penalty term. The second issue can be handled with the matrix factorization model, which is however non-convex. Besides, PCA based clustering can also be enhanced by using a graph of data similarity. In this article, we introduce a new model called 'Robust PCA on Graphs' which incorporates spectral graph regularization into the Robust PCA framework. Our proposed model benefits from 1) the robustness of principal components to occlusions and missing values, 2) enhanced low-rank recovery, 3) improved clustering property due to the graph smoothness assumption on the low-rank matrix, and 4) convexity of the resulting optimization problem. Extensive experiments on 8 benchmark, 3 video and 2 artificial datasets with corruptions clearly reveal that our model outperforms 10 other state-of-the-art models in its clustering and low-rank recovery tasks.

Convolutional Neural Networks (ConvNets) have successfully contributed to improve the accuracy of regression-based methods for computer vision tasks such as human pose estimation, landmark localization, and object detection. The network optimization has been usually performed with L2 loss and without considering the impact of outliers on the training process, where an outlier in this context is defined by a sample estimation that lies at an abnormal distance from the other training sample estimations in the objective space. In this work, we propose a regression model with ConvNets that achieves robustness to such outliers by minimizing Tukey's biweight function, an M-estimator robust to outliers, as the loss function for the ConvNet. In addition to the robust loss, we introduce a coarse-to-fine model, which processes input images of progressively higher resolutions for improving the accuracy of the regressed values. In our experiments, we demonstrate faster convergence and better generalization of our robust loss function for the tasks of human pose estimation and age estimation from face images. We also show that the combination of the robust loss function with the coarse-to-fine model produces comparable or better results than current state-of-the-art approaches in four publicly available human pose estimation datasets.

We present a new motion tracking method to robustly reconstruct non-rigid geometries and motions from single view depth inputs captured by a consumer depth sensor. The idea comes from the observation of the existence of intrinsic articulated subspace in most of non-rigid motions. To take advantage of this characteristic, we propose a novel L0 based motion regularizer with an iterative optimization solver that can implicitly constrain local deformation only on joints with articulated motions, leading to reduced solution space and physical plausible deformations. The L0 strategy is integrated into the available non-rigid motion tracking pipeline, forming the proposed L0-L2 non-rigid motion tracking method that can adaptively stop the tracking error propagation. Extensive experiments over complex human body motions with occlusions, face and hand motions demonstrate that our approach substantially improves tracking robustness and surface reconstruction accuracy.

We introduce a method for accurate three dimensional head pose estimation using a commodity depth camera. We perform pose estimation by registering a morphable face model to the measured depth data, using a combination of particle swarm optimization (PSO) and the iterative closest point (ICP) algorithm, which minimizes a cost function that includes a 3D registration and a 2D overlap term. The pose is estimated on the fly without requiring an explicit initialization or training phase. Our method handles large pose angles and partial occlusions by dynamically adapting to the reliable visible parts of the face. It is robust and generalizes to different depth sensors without modification. On the Biwi Kinect dataset, we achieve best-in-class performance, with average angular errors of 2.1, 2.1 and 2.4 degrees for yaw, pitch, and roll, respectively, and an average translational error of 5.9 mm, while running at 6 fps on a graphics processing unit.

The contour-guided color palette (CCP) is proposed for robust image segmentation. It efficiently integrates contour and color cues of an image. To find representative colors of an image, color samples along long contours between regions, similar in spirit to machine learning methodology that focus on samples near decision boundaries, are collected followed by the mean-shift (MS) algorithm in the sampled color space to achieve an image-dependent color palette. This color palette provides a preliminary segmentation in the spatial domain, which is further fine-tuned by post-processing techniques such as leakage avoidance, fake boundary removal, and small region mergence. Segmentation performances of CCP and MS are compared and analyzed. While CCP offers an acceptable standalone segmentation result, it can be further integrated into the framework of layered spectral segmentation to produce a more robust segmentation. The superior performance of CCP-based segmentation algorithm is demonstrated by experiments on the Berkeley Segmentation Dataset.

The ability to remotely measure heart rate from videos without requiring any special setup is beneficial to many applications. In recent years, a number of papers on heart rate (HR) measurement from videos have been proposed. However, these methods typically require the human subject to be stationary and for the illumination to be controlled. For methods that do take into account motion and illumination changes, strong assumptions are still made about the environment (e.g. background can be used for illumination rectification). In this paper, we propose an HR measurement method that is robust to motion, illumination changes, and does not require use of an environment's background. We present conditions under which cardiac activity extraction from local regions of the face can be treated as a linear Blind Source Separation problem and propose a simple but robust algorithm for selecting good local regions. The independent HR estimates from multiple local regions are then combined in a majority voting scheme that robustly recovers the HR. We validate our algorithm on a large database of challenging videos.

There have been tremendous improvements for facial landmark detection on general "in-the-wild" images. However, it is still challenging to detect the facial landmarks on images with severe occlusion and images with large head poses (e.g. profile face). In fact, the existing algorithms usually can only handle one of them. In this work, we propose a unified robust cascade regression framework that can handle both images with severe occlusion and images with large head poses. Specifically, the method iteratively predicts the landmark occlusions and the landmark locations. For occlusion estimation, instead of directly predicting the binary occlusion vectors, we introduce a supervised regression method that gradually updates the landmark visibility probabilities in each iteration to achieve robustness. In addition, we explicitly add occlusion pattern as a constraint to improve the performance of occlusion prediction. For landmark detection, we combine the landmark visibility probabilities, the local appearances, and the local shapes to iteratively update their positions. The experimental results show that the proposed method is significantly better than state-of-the-art works on images with severe occlusion and images with large head poses. It is also comparable to other methods on general "in-the-wild" images.

In perspective cameras, images of a frontal-parallel 3D object preserve its aspect ratio invariant to its depth. Such an invariance is useful in photography but is unique to perspective projection. In this paper, we show that alternative non-perspective cameras such as the crossed-slit or XSlit cameras exhibit a different depth-dependent aspect ratio (DDAR) property that can be used to 3D recovery. We first conduct a comprehensive analysis to characterize DDAR, infer object depth from its AR, and model recoverable depth range, sensitivity, and error. We show that repeated shape patterns in real Manhattan World scenes can be used for 3D reconstruction using a single XSlit image. We also extend our analysis to model slopes of lines. Specifically, parallel 3D lines exhibit depth-dependent slopes (DDS) on their images which can also be used to infer their depths. We validate our analyses using real XSlit cameras, XSlit panoramas, and catadioptric mirrors. Experiments show that DDAR and DDS provide important depth cues and enable effective single-image scene reconstruction.

Images of the eye are key in several computer vision problems, such as shape registration and gaze estimation. Recent large-scale supervised methods for these problems require time-consuming data collection and manual annotation, which can be unreliable. We propose synthesizing perfectly labelled photo-realistic training data in a fraction of the time. We used computer graphics techniques to build a collection of dynamic eye-region models from head scan geometry. These were randomly posed to synthesize close-up eye images for a wide range of head poses, gaze directions, and illumination conditions. We used our model's controllability to verify the importance of realistic illumination and shape variations in eye-region training data. Finally, we demonstrate the benefits of our synthesized training data (SynthesEyes) by out-performing state-of-the-art methods for eye-shape registration as well as cross-dataset appearance-based gaze estimation in the wild.

The development of deep learning has empowered machines with comparable capability of recognizing limited image categories to human beings. However, most existing approaches heavily rely on human-curated training data, which hinders the scalability to large and unlabeled vocabularies in image tagging. In this paper, we propose a weakly-supervised deep learning model which can be trained from the readily available Web images to relax the dependence on human labors and scale up to arbitrary tags (categories). Specifically, based on the assumption that features of true samples in a category tend to be similar and noises tend to be variant, we embed the feature map of the last deep layer into a new affinity representation, and further minimize the discrepancy between the affinity representation and its low-rank approximation. The discrepancy is finally transformed into the objective function to give relevance feedback to back propagation. Experiments show that we can achieve a performance gain of 14.0% in terms of a semantic-based relevance metric in image tagging with 63,043 tags from the WordNet, against the typical deep model trained on the ImageNet 1,000 vocabulary set.