Mingtao Pei

Vehicle Type Classification Using a Semisupervised Convolutional Neural Network

In this paper, we propose a vehicle type classification method using a semisupervised convolutional neural network from vehicle frontal-view images. In order to capture rich and discriminative information of vehicles, we introduce sparse Laplacian filter learning to obtain the filters of the network with large amounts of unlabeled data. Serving as the output layer of the network, the softmax classifier is trained by multitask learning with small amounts of labeled data. For a given vehicle image, the network can provide the probability of each type to which the vehicle belongs. Unlike traditional methods by using handcrafted visual features, our method is able to automatically learn good features for the classification task. The learned features are discriminative enough to work well in complex scenes. We build the challenging BIT-Vehicle dataset, including 9850 high-resolution vehicle frontal-view images. Experimental results on our own dataset and a public dataset demonstrate the effectiveness of the proposed method.

Robust Discriminative Tracking via Landmark-Based Label Propagation

The appearance of an object could be continuously changing during tracking, thereby being not independent identically distributed. A good discriminative tracker often needs a large number of training samples to fit the underlying data distribution, which is impractical for visual tracking. In this paper, we present a new discriminative tracker via landmark-based label propagation (LLP) that is nonparametric and makes no specific assumption about the sample distribution. With an undirected graph representation of samples, the LLP locally approximates the soft label of each sample by a linear combination of labels on its nearby landmarks. It is able to effectively propagate a limited amount of initial labels to a large amount of unlabeled samples. To this end, we introduce a local landmarks approximation method to compute the cross-similarity matrix between the whole data and landmarks. Moreover, a soft label prediction function incorporating the graph Laplacian regularizer is used to diffuse the known labels to all the unlabeled vertices in the graph, which explicitly considers the local geometrical structure of all samples. Tracking is then carried out within a Bayesian inference framework, where the soft label prediction value is used to construct the observation model. Both qualitative and quantitative evaluations on the benchmark data set containing 51 challenging image sequences demonstrate that the proposed algorithm outperforms the state-of-the-art methods.

Vehicle Type Classification Using Unsupervised Convolutional Neural Network

In this paper, we propose an appearance-based vehicle type classification method from vehicle frontal view images. Unlike other methods using hand-crafted visual features, our method is able to automatically learn good features for vehicle type classification by using a convolutional neural network. In order to capture rich and discriminative information of vehicles, the network is pre-trained by the sparse filtering which is an unsupervised learning method. Besides, the network is with layer-skipping to ensure that final features contain both high-level global and low-level local features. After the final features are obtained, the soft max regression is used to classify vehicle types. We build a challenging vehicle dataset called BIT-Vehicle dataset to evaluate the performance of our method. Experimental results on a public dataset and our own dataset demonstrate that our method is quite effective in classifying vehicle types.

Nonnegative Correlation Coding for Image Classification

Feature coding is one of the most important procedures in the bag-of-features model for image classification. In this paper, we propose a novel feature coding method called nonnegative correlation coding. In order to obtain a discriminative image representation, our method employs two correlations: the correlation between features and visual words, and the correlation between the obtained codes. The first correlation reflects the locality of codes, i.e., the visual words close to the local feature are activated more easily than the ones distant. The second correlation characterizes the similarity of codes, and it means that similar local features are likely to have similar codes. Both correlations are modeled under the nonnegative constraint. Based on the Nesterov’s gradient projection algorithm, we develop an effective numerical solver to optimize the nonnegative correlation coding problem with guaranteed quadratic convergence. Comprehensive experimental results on publicly available datasets demonstrate the effectiveness of our method.创新点本文提出了一种用于图像分类的编码方法,称为“非负关联编码”。为了获得有判别力的图像表示,非负关联编码利用了两种关系:一是待编码的局部特征与视觉单词之间的关系,它反映了编码过程的局部性,即局部特征倾向于利用距离它较近的视觉单词进行表达;二是编码之间的关系,它体现了编码过程的相似性,即相似的局部特征具有相似的编码。这两种关系都在非负约束的条件下建模。另外,本文基于NGP(Nesterov梯度投影)方法提出了一种用于求解非负关联编码的有效算法。公共数据集上的实验结果证明了方法的有效性。

Learning online structural appearance model for robust object tracking

The main challenge of robust object tracking comes from the difficulty in designing an adaptive appearance model that is able to accommodate appearance variations. Existing tracking algorithms often perform self-updating of the appearance model with examples from recent tracking results to account for appearance changes. However, slight inaccuracy of tracking results can degrade the appearance model. In this paper, we propose a robust tracking method by evaluating an online structural appearance model based on local sparse coding and online metric learning. Our appearance model employs pooling of structural features over the local sparse codes of an object region to obtain a middle-level object representation. Tracking is then formulated by seeking for the most similar candidate within a Bayesian inference framework where the distance metric for similarity measurement is learned in an online manner to match the varying object appearance. Both qualitative and quantitative evaluations on various challenging image sequences demonstrate that the proposed algorithm outperforms the state-of-the-art methods.

Coupling-and-decoupling: A hierarchical model for occlusion-free object detection

Abstract Handling occlusion is a very challenging problem in object detection. This paper presents a method of learning a hierarchical model for X-to-X occlusion-free object detection (e.g., car-to-car and person-to-person occlusions in our experiments). The proposed method is motivated by an intuitive coupling-and-decoupling strategy. In the learning stage, the pair of occluding X׳s (e.g., car pairs or person pairs) is represented directly and jointly by a hierarchical And–Or directed acyclic graph (AOG) which accounts for the statistically significant co-occurrence (i.e., coupling). The structure and the parameters of the AOG are learned using the latent structural SVM (LSSVM) framework. In detection, a dynamic programming (DP) algorithm is utilized to find the best parse trees for all sliding windows with detection scores being greater than the learned threshold. Then, the two single X׳s are decoupled from the declared detections of X-to-X occluding pairs together with some non-maximum suppression (NMS) post-processing. In experiments, our method is tested on both a roadside-car dataset collected by ourselves (which will be released with this paper) and two public person datasets, the MPII-2Person dataset and the TUD-Crossing dataset. Our method is compared with state-of-the-art deformable part-based methods, and obtains comparable or better detection performance.

Orthonormal dictionary learning and its application to face recognition

Abstract This paper presents an orthonormal dictionary learning method for low-rank representation. The orthonormal property encourages the dictionary atoms to be as dissimilar as possible, which is beneficial for reducing the ambiguities of representations and computation cost. To make the dictionary more discriminative, we enhance the ability of the class-specific dictionary to well represent samples from the associated class and suppress the ability of representing samples from other classes, and also enforce the representations that have small within-class scatter and big between-class scatter. The learned orthonormal dictionary is used to obtain low-rank representations with fast computation. The performances of face recognition demonstrate the effectiveness and efficiency of the method.

Robust object tracking via online multiple instance metric learning

This paper presents a novel object tracking method using online multiple instance metric learning to adaptively capture appearance variations. More specifically, we seek for an appropriate metric via online metric learning to match the different appearances of an object and simultaneously separate the object from the background. The drift problem caused by potentially misaligned training examples is alleviated by performing online metric learning under the multiple instance setting. Both qualitative and quantitative evaluations on various challenging sequences are discussed.

Online Discriminative Tracking With Active Example Selection

Most existing discriminative tracking algorithms use a sampling-and-labeling strategy to collect examples and treat the training example collection as a task that is independent of classifier learning. However, the examples collected directly by sampling are neither necessarily informative nor intended to be useful for classifier learning. Updating the classifier with these examples might introduce ambiguity to the tracker. In this paper, we present a novel online discriminative tracking framework that explicitly couples the objectives of example collection and classifier learning. Our method uses Laplacian regularized least squares (LapRLS) to learn a robust classifier that can sufficiently exploit unlabeled data and preserve the local geometrical structure of the feature space. To ensure the high classification confidence of the classifier, we propose an active example selection approach to automatically select the most informative examples for LapRLS. Part of the selected examples that satisfy strict constraints are labeled to enhance the adaptivity of our tracker, which actually provides robust supervisory information to guide semisupervised learning. With active example selection, we are able to avoid the ambiguity introduced by an independent example collection strategy and to alleviate the drift problem caused by misaligned examples. Comparison with the state-of-the-art trackers on the comprehensive benchmark demonstrates that our tracking algorithm is more effective and accurate.

Coupling Semi-supervised Learning and Example Selection for Online Object Tracking

Training example collection is of great importance for discriminative trackers. Most existing algorithms use a sampling-and-labeling strategy, and treat the training example collection as a task that is independent of classifier learning. However, the examples collected directly by sampling are not intended to be useful for classifier learning. Updating the classifier with these examples might introduce ambiguity to the tracker. In this paper, we introduce an active example selection stage between sampling and labeling, and propose a novel online object tracking algorithm which explicitly couples the objectives of semi-supervised learning and example selection. Our method uses Laplacian Regularized Least Squares (LapRLS) to learn a robust classifier that can sufficiently exploit unlabeled data and preserve the local geometrical structure of feature space. To ensure the high classification confidence of the classifier, we propose an active example selection approach to automatically select the most informative examples for LapRLS. Part of the selected examples that satisfy strict constraints are labeled to enhance the adaptivity of our tracker, which actually provides robust supervisory information to guide semi-supervised learning. With active example selection, we are able to avoid the ambiguity introduced by an independent example collection strategy, and to alleviate the drift problem caused by misaligned examples. Comparison with the state-of-the-art trackers on the comprehensive benchmark demonstrates that our tracking algorithm is more effective and accurate.