Computer Vision

	Project Details
SL No.	Topic Name
1	Tracking objects in videos with camera movement (Pan-Tilt-Zoom)
2	Action Recognition/classification from Video shots
3	Context based Video Summerization
4	Track and Representation of multiple moving objects (under occlusion) from video
5	Scale space approach for video analytics (Theoretical Study and Feature Descriptors)
6	Face Recognition using Degraded face image samples
7	Decision fusion of recent Face Recognizers under degradation, illumination, and pose variations (e.g. LTP, Dual-Space, SVM-based, K-DDA, K-DCV, LLE, CCA, Sparse-proj, K-LDA, subband-face )
8	Specularity Estimation from frontal Face Images – use for super-resolution and enhancement
9	Pose detection on a face image, using automatic identification of landmarks (say, using AAM )
10	Visual Saliency: a top-down Approach
11	Scale invariant multi-object detection using a set of classifiers (say, AdaBoost)
12	Supervised classification of foreground and background objects separately in a scene
13	SLAR (Simultaneous Localization And Recognition) frame work for part-based object recognition
14	Supervised Multi-cut algorithms for scene segmentation

Computer Graphics

	Project Details
SL No.	Topic Name
1	Realistic Human Face Rendering with Emotion, expression, hair-style and aging effects
2	Video based rendering of moving objects and 3D scenes
3	GDL (Game Description Language): API Design
4	Character Design: Animation, Soft Object Modeling, and Audio
5	GPU-based realistic and online rendering

Machine Learning Algorithms for Vision

	Project Details
SL No.	Topic Name
1	Domain Adaptive Feature Transform
2	Learning Saliency in Images/Video
3	Deep/Hard/Transfer Learning for Organizing/Classification of Large Databases
4	Recent Advances - Compressed Sensing, Sparse Representation, BOW, BOF
5	Matching Pursuit, ADMM and LASSO for Image Super-resolution, Edge-Preserving Contrast Enhancement
6	Proximal Gradient for Sparse Dictionary Learning from Few Samples - Applications: FR, Video Organization/Retrieval/Learning
7	Deep Learning Methods for (i) Event Recall from Video; (ii) Object Recognition from Complex Scenes in Presence of Clutter, Pose Variations etc.; (iii) FR and (iv) SLAR
8	Combining FIT, Deep Learning, Top-Down Saliency cum Attention, DA and Sparse Representation for Visual Interpretation

Pattern Recognition and Visual Perception

	Project Details
SL No.	Topic Name
1	Memory modules for face/Object recognition
2	Characterizing object shapes using features derived from top-points
3	Identifying similar and beautiful faces
4	Modeling Thoughts, emotion and ego