Manohar Paluri

Since I joined Facebook in 2012 I have worked on building Facebook’s Image and Video understanding platform. My interests lie in using large amounts of annotated and weakly annotated image and video data to build a visual representation of everything in the world. Prior to joining Facebook I was pursuing my PhD from Georgia Tech and worked at Google, IBM Watson labs and Sarnoff Corporation (now SRI). Currently, I manage the computer vision group and my group focuses on building Facebook’s image and video understanding platform.

Interests

Computer vision and machine learning

Website

Google Scholar Profile

Website

Google Scholar Profile

Interests

Latest Publications

Large-Scale Visual Relationship Understanding

Self-Supervised Feature Learning for Semantic Segmentation of Overhead Imagery

A Closer Look at Spatiotemporal Convolutions for Action Recognition

Detect-and-Track: Efficient Pose Estimation in Videos

What Makes a Video a Video: Analyzing Temporal Information in Video Understanding Models and Datasets

Exploring the Limits of Weakly Supervised Pretraining

Cultural Diffusion and Trends in Facebook Photographs

Unsupervised Learning of Edges

End-to-End Voxel-to-Voxel Prediction

Metric Learning with Adaptive Density Discrimination

Learning Spatiotemporal Features with 3D Convolutional Networks

Beyond Frontal Faces: Improving Person Recognition Using Multiple Cues

PANDA: Pose Aligned Networks for Deep Attribute Modeling

Show More Publications

News

Advancing state-of-the-art image recognition with deep learning on hashtags

Announcing the DeepGlobe Satellite Challenge for CVPR 2018

Engineering for nostalgia: Building a personalized ‘On This Day’ experience

C3D: Generic Features for Video Analysis

Show More News