Research Interest

I am interested in Computer Vision and Machine Learning. At present, I specifically focus my research on probabilistic object tracking and general object recognition.

Thanks to large-scale image repositories, vast amounts of data for object recognition are now easily available. However, acquiring training labels for arbitrary objects still requires tedious and expensive human effort. This is particularly true for localization, where humans must not only provide labels, but also training windows in an image. We present an approach for reducing the number of labelled training instances required to train an object classifier and for assisting the user in specifying optimal object location windows. As part of this process, the algorithm performs localization to find bounding windows for training examples that are best aligned with the current classification function, which optimizes learning and reduces human effort. To test this approach, we introduce an active learning extension to a latent SVM learning algorithm. Our user interface for training object detectors employs real-time interaction with a human user. Our active learning system provides a mean performance improvement of 4.5% in the average precision over a state of the art detector on the PASCAL Visual Object Classes Challenge 2007 with an average of just 40 minutes of human labelling effort per class.

Demo video of our interface: [mpeg in color (3.6MB)]
Presentation: [PDF (4.5MB)]

Boosted Particle Filter is a vision system that is capable of learning, detecting and tracking the objects of interest.
The problem of tracking a varying number of non-rigid objects has two major difficulties. First, the observation models and target distributions can be highly non-linear and non-Gaussian. Second, the presence of a large, varying number of objects creates complex interactions with overlap and ambiguities. To surmount these difficulties, we introduce BPF.
The system is demonstrated in the context of tracking hockey players using video sequences. Our approach combines the strengths of two successful algorithms: mixture particle filters and Adaboost. The mixture particle filter is ideally suited to multi-target tracking as it assigns a mixture component to each player. The crucial design issues in mixture particle filters are the choice of the proposal distribution and the treatment of objects leaving and entering the scene. With BPF, we construct the proposal distribution using a mixture model that incorporates information from the dynamic models of each player and the detection hypotheses generated by Adaboost. The learned Adaboost proposal distribution allows us to quickly detect players entering the scene, while the filtering process enables us to keep track of the individual players. The result of interleaving Adaboost with mixture particle filters is a simple, yet powerful and fully automatic multiple object tracking system.

Movies: [mpeg in color (838KB)] [mpeg in gray scale (343KB)]
Presentation: [mpeg (3.5MB)]
Source code (matlab) and data (jpegs): [ Ver. 1.3 (zip, 3.3MB)]

Objective:
Development of the system that automatically acquires trajectories of hockey players from video.

This project has two main stages:

1. Compute the mapping from video to the rink map

&

||

We choose 13 point correspondences manually from the original image and the rink map and compute homography between them. The result of the transformation is shown above. Using this transformation for the first frame of a sequence, we now automatically compute transformations for the rest of the sequence. We implement KLT, RANSAC, and model fitting to do this automation.

Two video clips (about 800 frames each) are presented below. One is a clip of the original video. The other is a clip of a tranformed sequence.

NOTE: Above transformed sequence looks like a fast-forward version of the original sequence transformed. This is because we process the sequence every 4th frames.

2. Track hockey players

Now that we aquire the transformation from the original sequence to the rink map, we track hockey players in order to gain their trajectories.

Here is the example clip of how a hockey player is tracked:

Movie:

We finally combine all the results to visualize trajectories of hockey players.

The trajectory of the movie above.(Only a single player for now):