CVIR IIT Kharagpur

This is the laboratory webpage of Computer Vision and Intelligence Research Group at Department of Computer Science and Engineering, IIT Kharagpur, India, led by Prof. Abir Das.

News

[Feb'24] Convolutional Prompting meets Language Models for Continual Learning accepted at CVPR 2024.
[Aug'23] Exemplar-Free Continual Transformer with Convolutions accepted at ICCV 2023.
[Jan'23] Select, Label, and Mix (SLM) received the Best Paper Honorable Mention Award at WACV 2023!.
[Oct'22] Paper on Partial Domain Adaptation accepted at WACV 2023.
[Sept'21] Paper on Domain Adaptation in Action Recognition accepted at NeurIPS 2021.
[Feb'21] Paper on Semi-Supervised Action Recognition accepted at CVPR 2021.

Research

Publications

Convolutional Prompting meets Language Models for Continual Learning
Anurag Roy, Riddhiman Moulick, Vinay K. Verma, Saptarshi Ghosh, Abir Das
Computer Vision and Pattern Recognition (CVPR), 2024.
project page / pdf / code

We develop a parameter and compute efficient approach called ConvPrompt that leverages convolutional prompt creation and Large Language Models for enhanced knowledge sharing and concept transfer in Continual Learning.

Exemplar-Free Continual Transformer with Convolutions
Anurag Roy, Vinay K. Verma, Sravan Voonna, Kripabandhu Ghosh, Saptarshi Ghosh, Abir Das
International Conference on Computer Vision (ICCV), 2023.
project page / pdf / code

We develop a novel approach called ConTraCon that leverages convolutions on transformer weights for continual learning without requiring task identifiers or storing examples from previous tasks.

Select, Label, and Mix: Learning Discriminative Invariant Feature Representations for Partial Domain Adaptation
Aadarsh Sahoo, Rameswar Panda, Rogerio Feris, Kate Saenko, Abir Das
NeurIPS DistShift Workshop (NeurIPS-W), 2021.
Winter Conference on Applications of Computer Vision (WACV), 2023.
(Best Paper Honorable Mention).
project page / poster / video presentation / slides / code

We develop a novel 'Select, Label, and Mix' (SLM) framework that aims to learn discriminative invariant feature representations for partial domain adaptation.

Contrast and Mix: Temporal Contrastive Video Domain Adaptation with Background Mixing
Aadarsh Sahoo, Rutav Shah, Rameswar Panda, Kate Saenko, Abir Das
35th Conference on Neural Information Processing Systems (NeurIPS), 2021.
project page / poster / video presentation / slides / code

We introduce a novel temporal contrastive learning approach for unsupervised video domain adaptation, which is achieved by jointly leveraging video speed, background mixing, and target pseudo-labels.

Reinforcement Explanation Learning
Siddhant Agarwal, Owais Iqbal, Sree Aditya Buridi, Mada Manjusha, Abir Das
35th Conference on Neural Information Processing Systems (NeurIPS), 2021.
project page / arXiv / code

We reformulate the process of generating saliency maps using perturbation based methods for black box models as a Markov Decsion Process and use RL to optimally search for the best saliency map, thereby reducing the inference time without hurting the performance.

Semi-Supervised Action Recognition with Temporal Contrastive Learning
Ankit Singh, Omprakash Chakraborty, Ashutosh Varshney, Rameswar Panda, Rogerio Feris, Kate Saenko, Abir Das
Computer Vision and Pattern Recognition (CVPR), 2021.
project page / poster / video presentation / code

We address semi-supervised video action recognition by learning a two-pathway temporal contrastive model where the similarity between representations of the same video at two different speeds are maximized while the similarity between different videos played at different speeds are minimized.

Webpage template courtesy: Jon Barron