About

I am currently a senior research scientist at Google DeepMind, working on computer vision and machine learning. Prior to that I completed my PhD from UC Riverside, advised by Amit K Roy-Chowdhury. During my PhD I also spend three summers at NEC Labs, MERL and Microsoft Research.

Research Interests

My research interests lie in learning problems in computer vision and robotics using limited human supervision. In the long run, I am interested in developing algorithms which can be learned with low supervision, be generalizable, robust, just like humans, and help in our everyday life.

Selected Publications

Mixture of Nested Experts: Adaptive Processing of Visual Tokens, NeurIPS '24

Gagan Jain*, Nidhi Hegde, Aditya Kusupati, Arsha Nagrani, Shyamal Buch, Prateek Jain, Anurag Arnab, Sujoy Paul*

LookupViT: Compressing visual information to a limited number of tokens, ECCV '24

Rajat Koner, Gagan Jain, Prateek Jain, Volker Tresp, Sujoy Paul

Test-time adaptation with slot-centric models, ICML '23

Mihir Prabhudesai, Sujoy Paul, Sjoerd van Steenkiste, Mehdi Sajjadi, Anirudh Goyal, Deepak Pathak, Katerina Fragkiadaki, Gaurav Aggarwal, Thomas Kipf

Weakly supervised information extraction from inscrutable handwritten document images, ICDAR '23

Sujoy Paul, Gagan Madan, Akankshya Mishra, Narayan Hegde, Pradeep Kumar, Gaurav Aggarwal

SITA: Single Image Test-time Adaptation, CVPR-W '23

Ansh Khurana, Sujoy Paul, Piyush Rai, Soma Biswas, Gaurav Aggarwal

Novel Class Discovery without Forgetting, ECCV '22

KJ Joseph, Sujoy Paul, Gaurav Aggarwal, Soma Biswas, Piyush Rai, Kai Han, Vineeth Balasubramanian

Unsupervised Adaptation of Semantic Segmentation Models without Source Data, ICML-W '22

Sujoy Paul, Ansh Khurana, Gaurav Aggarwal

Cross-domain Imitation from Observations, ICML '21 Oral

Dripta Raychaudhuri*, Sujoy Paul*, Jeroen van Baar, Amit Roy-Chowdhury

Unsupervised Multi-source Domain Adaptation Without Access to Source Data, CVPR '21 Oral

Miraj Ahmed*, Dripta Raychaudhuri*, Sujoy Paul*, Samet Oymak, Amit Roy-Chowdhury

Adversarial Knowledge Transfer from Unlabeled Data, ACM-MM '20

Akash Gupta, Rameswar Panda, Sujoy Paul, Jianming Zhang, Amit Roy-Chowdhury

Domain Adaptive Semantic Segmentation Using Weak Labels, ECCV '20

Sujoy Paul, Yi-Hsuan Tsai, Samuel Schulter, Amit Roy-Chowdhury, Manmohan Chandraker

Learning from Trajectories via Subgoal Discovery, NeurIPS '19

Sujoy Paul, Jeroen van Baar, Amit Roy-Chowdhury

Weakly Supervised Video Moment Retrieval From Text Queries, CVPR '19

Niluthpol Mithun*, Sujoy Paul*, Amit Roy-Chowdhury

Context-Aware Query Selection for Active Learning in Event Recognition, PAMI '18

Mahmudul Hasan*, Sujoy Paul*, Anastasios I. Mourikis, Amit Roy-Chowdhury

W-TALC: Weakly-supervised Temporal Activity Localization and Classification, ECCV '18

Sujoy Paul, Sourya Roy, Amit Roy-Chowdhury

Exploiting Transitivity for Learning Person Re-identification Models on a Budget, CVPR '18

Sourya Roy, Sujoy Paul, Neal E. Young, Amit Roy-Chowdhury

Non-Uniform Subset Selection for Active Learning in Structured Data, CVPR '17

Sujoy Paul, Sourya Roy, Amit Roy-Chowdhury

Online Adaptation for Joint Scene and Object Classification, ECCV '16

Jawad H. Bappy, Sujoy Paul, Amit K. Roy-Chowdhury