Event Location: NSH 1507
Bio: I am a fifth year graduate student with Prof. Jitendra Malik in the Vision group at University of California Berkeley. I did my undergraduate at the Indian Institute of Technology, Delhi. My interests are in Computer Vision and Machine learning. I am funded by a Microsoft Research Fellowship.

Abstract: Object recognition in computer vision comes in many flavors, two of the most popular being object detection and semantic segmentation. Object detection systems detect every instance of a category in an image, and coarsely localize each with a bounding box. Semantic segmentation systems assign category labels to pixels, thus providing pixel-precise localization but failing to resolve individual instances of the category. In this talk I argue that we should aim for a richer output: recognition systems should detect individual instances of a category and provide pixel precise segmentations for each. I will describe our approach for tackling this task (which we call simultaneous detection and segmentation) and show how one can leverage convolutional neural networks for precise localization. I will also show that the methods I describe are also effective for segmenting parts and localizing keypoints. These are our first steps towards a unified system that detects objects, segments them out and estimates their pose.