I am an upcoming doctoral student at the University of Texas at Austin, under the supervision of Dr. Georgios Pavlakos.
I have worked on various projects in the fields of Pose Classification and Contact Estimation, Tiny Object Detection and Classification, Audio Classification, and Biomedical Image Segmentation, under the purview of Computer Vision. I am open to working on new projects on topics both known and unexplored by me.
I developed a discrete contact annotation tool for vertex-level contact annotation, under Dr. Michael Black. I also created a vertex-level human contact dataset and developed a framework for the detection of contact from natural images of humans. The work has been accepted at ICCV 2023.
Currently, I'm working on developing an HPS methodology to extract accurate and realistic poses at scale from images and videos.
I developed a combined detection-classification pipeline for High-Resolution images of Trichogramma wasps. The work has been published at VISAPP 2023.
I developed a framework for the diagnosis of patients as ALS/PD or Normal, based on phoneme utterance audios, under Dr. Prasanta Ghosh.
I developed a framework for Automatic Pose Identification and Recommendation for Yoga asanas, under Dr. Brejesh Lall.
I performed independent research under Dr. Debangshu Dey on Glaucoma detection using deep learning methods. The submitted manuscript has been published in Biomedical Signal Processing and Control, Elsevier.
CGPA: 8.7/10.0
Proposed a pipeline to infer dense 3D contact on the human body using scene and body-part context.
Also curated a dataset with in-the-wild images and crowdsourced dense 3D contact annotations.
It has been accepted at the International Conference on Computer Vision (ICCV), 2023 (Oral).
Proposed a combined detection-classification pipeline for the detection of tiny wasps from images, and subsequent classification into species.
It has been accepted for oral presentation at the International Conference on Computer Vision Theory and Applications (VISAPP), 2023.
The proposed method leverages segmentation of the interpolation of two unlabeled data for Semi-Supervised Cardiac MRI segmentation.
It has been accepted for publication at IEEE ISBI, 2022.
A pipeline with auxiliary and densely-connected pyramidal decoder is proposed for segmentation of the Optic Disc and Cup from Optical Fundus images, and the subsequent classification into glaucomatous or not.
It has been accepted for publication at Biomedical Signal Processing and Control, Elsevier.
