We introduce a new representation learning approach for domain adaptation, in which data at training and test time come from similar but different distributions. Our approach is directly inspired by the theory on domain adaptation suggesting that, for effective domain transfer to be achieved, predictions must be made based on features that cannot discriminate between the training (source) and test (target) domains.

The approach implements this idea in the context of neural network architectures that are trained on labeled data from the source domain and unlabeled data from the target domain (no labeled target-domain data is necessary). As the training progresses, the approach promotes the emergence of features that are (i) discriminative for the main learning task on the source domain and (ii) indiscriminate with respect to the shift between the domains. We show that this adaptation behaviour can be achieved in almost any feed-forward model by augmenting it with few standard layers and a new gradient reversal layer. The resulting augmented architecture can be trained using standard backpropagation and stochastic gradient descent, and can thus be implemented with little effort using any of the deep learning packages.

We demonstrate the success of our approach for two distinct classification problems (document sentiment analysis and image classification), where state-of-the-art domain adaptation performance on standard benchmarks is achieved. We also validate the approach for descriptor learning task in the context of person re-identification application.

/pdf/domain-adversarial-training-of-neural-networks-53puwk7wkk.pdf

Domain-adversarial training of neural networks

Object tracking is one of the most important components in numerous applications of computer vision. While much progress has been made in recent years with efforts on sharing code and datasets, it is of great importance to develop a library and benchmark to gauge the state of the art. After briefly reviewing recent advances of online object tracking, we carry out large scale experiments with various evaluation criteria to understand how these algorithms perform. The test image sequences are annotated with different attributes for performance evaluation and analysis. By analyzing quantitative results, we identify effective approaches for robust tracking and provide potential future research directions in this field.

/pdf/online-object-tracking-a-benchmark-10et4l3bf2.pdf

Online Object Tracking: A Benchmark

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.

/pdf/scalable-person-re-identification-a-benchmark-iiee63lja2.pdf

Scalable Person Re-identification: A Benchmark

This paper investigates long-term tracking of unknown objects in a video stream. The object is defined by its location and extent in a single frame. In every frame that follows, the task is to determine the object's location and extent or indicate that the object is not present. We propose a novel tracking framework (TLD) that explicitly decomposes the long-term tracking task into tracking, learning, and detection. The tracker follows the object from frame to frame. The detector localizes all appearances that have been observed so far and corrects the tracker if necessary. The learning estimates the detector's errors and updates it to avoid these errors in the future. We study how to identify the detector's errors and learn from them. We develop a novel learning method (P-N learning) which estimates the errors by a pair of “experts”: (1) P-expert estimates missed detections, and (2) N-expert estimates false alarms. The learning process is modeled as a discrete dynamical system and the conditions under which the learning guarantees improvement are found. We describe our real-time implementation of the TLD framework and the P-N learning. We carry out an extensive quantitative evaluation which shows a significant improvement over state-of-the-art approaches.

Tracking-Learning-Detection

Object tracking has been one of the most important and active research areas in the field of computer vision. A large number of tracking algorithms have been proposed in recent years with demonstrated success. However, the set of sequences used for evaluation is often not sufficient or is sometimes biased for certain types of algorithms. Many datasets do not have common ground-truth object positions or extents, and this makes comparisons among the reported quantitative results difficult. In addition, the initial conditions or parameters of the evaluated tracking algorithms are not the same, and thus, the quantitative results reported in literature are incomparable or sometimes contradictory. To address these issues, we carry out an extensive evaluation of the state-of-the-art online object-tracking algorithms with various evaluation criteria to understand how these methods perform within the same framework. In this work, we first construct a large dataset with ground-truth object positions and extents for tracking and introduce the sequence attributes for the performance analysis. Second, we integrate most of the publicly available trackers into one code library with uniform input and output formats to facilitate large-scale performance evaluation. Third, we extensively evaluate the performance of 31 algorithms on 100 sequences with different initialization settings. By analyzing the quantitative results, we identify effective approaches for robust tracking and provide potential future research directions in this field.

/pdf/object-tracking-benchmark-o3bpdhhb3z.pdf

Object Tracking Benchmark

Traditionally, appearance models for recognition, reacquisition and tracking problems have been evaluated independently using metrics applied to a complete system. It is shown that appearance models for these three problems can be evaluated using a cumulative matching curve on a standardized dataset, and that this one curve can be converted to a synthetic reacquisition or disambiguation rate for tracking. A challenging new dataset for viewpoint invariant pedestrian recognition (VIPeR) is provided as an example. This dataset contains 632 pedestrian image pairs from arbitrary viewpoints. Several baseline methods are tested on this dataset and the results are presented as a benchmark for future appearance models and matchin methods.

Evaluating Appearance Models for Recognition, Reacquisition, and Tracking

This paper treats tracking as a foreground/background classification problem and proposes an online semi- supervised learning framework. Initialized with a small number of labeled samples, semi-supervised learning treats each new sample as unlabeled data. Classification of new data and updating of the classifier are achieved simultaneously in a co-training framework. The object is represented using independent features and an online support vector machine (SVM) is built for each feature. The predictions from different features are fused by combining the confidence map from each classifier using a classifier weighting method which creates a final classifier that performs better than any classifier based on a single feature. The semi-supervised learning approach then uses the output of the combined confidence map to generate new samples and update the SVMs online. With this approach, the tracker gains increasing knowledge of the object and background and continually improves itself over time. Compared to other discriminative trackers, the online semi-supervised learning approach improves each individual classifier using the information from other features, thus leading to a more robust tracker. Experiments show that this framework performs better than state-of-the-art tracking algorithms on challenging sequences.

/pdf/co-tracking-using-semi-supervised-support-vector-machines-dm0celtlt3.pdf

Co-Tracking Using Semi-Supervised Support Vector Machines

Illumination changes cause object appearance to change drastically and many existing tracking algorithms lack the capability to handle this problem. The Earth mover's distance (EMD) is a similarity measure that is more robust against illumination changes. However, EMD is computationally expensive and we therefore propose the differential EMD (DEMD) algorithm which computes the derivative of the EMD with respect to the object location so that the EMD does not need to be computed for every location in the tracking window. The fast differential formula is derived based on the sensitivity analysis of the simplex method as applied to the EMD formula. To further reduce the computation, signatures, i.e., variable-size descriptions of distributions, are employed as an object representation. The new algorithm models local background scenes as well as foreground objects to handle scale changes in a principled way. Extensive quantitative evaluation of the proposed algorithm has been carried out using benchmark sequences and the improvement over the standard mean shift tracker is demonstrated.

Differential EMD Tracking

Image inpainting is an old technique for filling in missing or damaged regions of an image. The general problem is: given an image and a user-defined region of the image to be filled, fill the region in such a manner that one cannot tell that the region has been altered. In this project I have implemented the method proposed by Bertalmio et al in their 2003 paper “Simultaneous Structure and Texture Image Inpainting” to accomplish this task automatically on digital images.

/pdf/ee264-project-simultaneous-structure-and-texture-image-lk37xf7wef.pdf

Shane Brennan

Papers

Evaluating Appearance Models for Recognition, Reacquisition, and Tracking

Co-Tracking Using Semi-Supervised Support Vector Machines

Differential EMD Tracking

EE264 Project : Simultaneous Structure and Texture Image Inpainting