Behjat Siddiquie

Amazon, Seattle, WA
behjats at gmail dot com
I am a Senior Research Scientist at Amazon, working on Amazon Go, a no checkout grocery store. We are using Computer Vision, Machine Learning and Deep Learning to re-invent retail.

Previously, I was a Senior Computer Scientist at SRI International, where I worked on real-time human behavior recognition and interaction modeling using multimodal sensors, and on using image analytics to enhance social media based security applications.

I completed my PhD from the University of Maryland, College Park, under the guidance of Prof. Larry Davis. In my thesis, I addressed the problem of efficiently retrieving images from large databases based on complex and multi-modal queries. Before that, I completed my BTech. from the Indian Institute of Technology, Bombay.

Selected Publications (Complete List)

Google Scholar Citations
  • Energy-Based Learning for Scene Graph Generation
    IEEE Conference on Computer Vision and Pattern Recognition, (CVPR) 2021, (nominated for the best paper award)
    Mohammed Suhail, Abhay Mittal, Behjat Siddiquie, Chris Broaddus, Jayan Eledath, Gerard Medioni and Leonid Sigal
    [paper][code][video]

  • Deep Multimodal Fusion: A Hybrid Approach
    International Journal of Computer Vision, (IJCV) 2018
    Mohamed Amer, Tim Shields, Behjat Siddiquie, Amir Tamrakar, Ajay Divakaran and Sek Chai
    [link]

  • Exploiting Multimodal Affect and Semantics to Identify Politically Persuasive Web Videos
    ACM International Conference on Multimodal Interaction, (ICMI) 2015
    Behjat Siddiquie, Dave Chisholm and Ajay Divakaran
    [pdf] [poster]

  • Multi-Modal Image Retrieval for Complex Queries using Small Codes
    ACM International Conference on Multimedia Retrieval, (ICMR) 2014
    Behjat Siddiquie, Brandyn White, Abhishek Sharma and Larry S. Davis
    [pdf] [poster] [Supplementary Material]

  • Emotion Detection in Speech using Deep Networks
    IEEE International Conference on Acoustics, Speech, and Signal Processing, (ICASSP) 2014
    Mohamed Amer, Behjat Siddiquie, Colleen Richey and Ajay Divakaran
    [pdf] [poster]

  • Large-Scale Vehicle Detection, Indexing, and Search in Urban Surveillance Videos
    IEEE Transactions on Multimedia, 2012
    Rogerio Feris, Behjat Siddiquie, James Petterson, Yun Zhai, Ankur Datta, Lisa Brown and Sharath Pankanti
    [link]

  • Image Ranking and Retrieval based on Multi-Attribute Queries
    IEEE Conference on Computer Vision and Pattern Recognition, (CVPR) 2011, (Oral Presentation)
    Behjat Siddiquie, Rogerio S. Feris and Larry S. Davis
    [pdf] [slides] [talk]

  • Beyond Active Noun Tagging: Modeling Contextual Interactions for Multi-Class Active Learning
    IEEE Conference on Computer Vision and Pattern Recognition, (CVPR) 2010, (Oral Presentation)
    Behjat Siddiquie and Abhinav Gupta
    [pdf] [slides] [poster] [talk]

  • Combining Multiple Kernels for Efficient Image Classification
    IEEE Workshop on Applications of Computer Vision, (WACV) 2009
    Behjat Siddiquie, Shiv N. Vitaladevuni and Larry S. Davis
    [pdf] [poster]

  • Incremental Multiple Kernel Learning for Object Recognition
    IEEE International Conference on Computer Vision, (ICCV) 2009
    Aniruddha Kembhavi, Behjat Siddiquie, Roland Miezianko, Scott McCloskey and Larry S. Davis
    [pdf] [poster] [video]

Patents

  • Exploiting multi-modal affect and semantics to assess the persuasiveness of a video (SRI International) [14/874,348]

  • Recognizing salient video events through learning-based multimodal analysis of visual features and audio-based analytics (SRI International) [14/846,318]

  • Dynamic hybrid models for multimodal analysis (SRI International) [9,875,445]

  • Multi-Modal Modeling of Temporal Interaction Sequences (SRI International) [9,734,730]

  • Multi-View Object Detection using Appearance Model Transfer from Similar Scenes (IBM) [8,498,448]

  • Video based Detection of Multiple Object Types under Varying Poses (IBM) [8,620,026]

  • Object Detection in Crowded Scenes (IBM) [8,811,663]

  • Image Ranking Based on Attribute Correlation (IBM) [9,262,445]