arXiv subjects

Jitendra Malik

Publications and source records attributed to Jitendra Malik.

255 records · Page 15Linked to original sources

arXiv2014

Pixels to Voxels: Modeling Visual Representation in the Human Brain

The human brain is adept at solving difficult high-level visual processing problems such as image interpretation and object recognition in natural scenes. Over the past few years neuroscientists have made remarkable progress in understanding how the human brain represents categories of objects and actions in natural scenes. However, all current models of high-level human vision operate on hand annotated images in which the objects and actions have been assigned semantic tags by a human operator. No current models can account for high-level visual function directly in terms of low-level visual input (i.e., pixels). To overcome this fundamental limitation we sought to develop a new class of models that can predict human brain activity directly from low-level visual input (i.e., pixels). We explored two classes of models drawn from computer vision and machine learning. The first class of models was based on Fisher Vectors (FV) and the second was based on Convolutional Neural Networks (ConvNets). We find that both classes of models accurately predict brain activity in high-level visual areas, directly from pixels and without the need for any semantic tags or hand annotation of images. This is the first time that such a mapping has been obtained. The fit models provide a new platform for exploring the functional principles of human vision, and they show that modern methods of computer vision and machine learning provide important tools for characterizing brain function.

q-bio.NC↗

arXiv2014

Simultaneous Detection and Segmentation

We aim to detect all instances of a category in an image and, for each instance, mark the pixels that belong to it. We call this task Simultaneous Detection and Segmentation (SDS). Unlike classical bounding box detection, SDS requires a segmentation and not just a box. Unlike classical semantic segmentation, we require individual object instances. We build on recent work that uses convolutional neural networks to classify category-independent region proposals (R-CNN [16]), introducing a novel architecture tailored for SDS. We then use category-specific, top- down figure-ground predictions to refine our bottom-up proposals. We show a 7 point boost (16% relative) over our baselines on SDS, a 5 point boost (10% relative) over state-of-the-art on semantic segmentation, and state-of-the-art performance in object detection. Finally, we provide diagnostic tools that unpack performance and provide directions for future work.

cs.CV↗

arXiv2014

R-CNNs for Pose Estimation and Action Detection

We present convolutional neural networks for the tasks of keypoint (pose) prediction and action classification of people in unconstrained images. Our approach involves training an R-CNN detector with loss functions depending on the task being tackled. We evaluate our method on the challenging PASCAL VOC dataset and compare it to previous leading approaches. Our method gives state-of-the-art results for keypoint and action prediction. Additionally, we introduce a new dataset for action detection, the task of simultaneously localizing people and classifying their actions, and present results using our approach.

cs.CV↗