Vishal Pani

I am a Master's student at EPFL primarily focussing on Computer Graphics and Vision applications.

During my Bachelor's at IIIT-Allahabad, I worked on Computer Vision related problems such as Visual Place Recognition (VPR), and Multi-Object Tracking (MOT) over a Long Horizon. Currently, as a Master's student at EPFL, I am working on the exciting domain of Differentiable Rendering with the Realistic Graphics Lab (RGL) led by Prof. Wenzel Jakob .

Here is my Resume.

Email  /  Github  /  Google Scholar  /  Linkedin

profile photo
Publications
TrichTrack: Multi-Object Tracking of Small-Scale Trichogramma Wasps

Vishal Pani, Martin Bernet, Vincent Calcagno, Louise Van Oudenhove, François Bremond
Advanced Video and Signal-based Surveillance (AVSS), 2021
Paper / Slides

Multi-Object Tracking with Re-Identification to study interactions between Trichogramma wasps over a long horizon. As part of my Bachelor's thesis at Inria, Sophia Antipolis under the supervision of Dr. François Bremond.

Research Projects
project image Gonio-photometer Calibration & Mesh Reconstruction using Differentiable Rendering Algorithms

Supervisor: Dr. Wenzel Jakob
More info on the gonio-photometer

Mesh reconstruction using Large steps in inverse rendering of geometry and accurate captures from a gonio-photometer setup.

project image On the Association of Textured Meshes and Imagery for Multi-Modal Semantic Segmentation

Supervisor: apl. Prof. Dr.-Ing. Norbert Haala
More info on the umbrella project

A complete data preparation (via ray casting) and deep learning pipeline for Semantic Image Segmentation of City Meshes using 2D Projections of the 3D Meshes.

project image Multi-Domain Localization of Autonomous Agent

Supervisor: Dr. Rahul Kala
Code

Localization of an agent running in a different domains by comparing with a static map. Siamese embeddings, custom dynamic-branched CNN, and particle filters used.

project image Stabilization Techniques for the Training of Generative Adversarial Networks

Supervisor: Dr. Nikhil R. Pal
Code

Stabilized the error oscillation of Discriminator by adding a damping term to its. objective function. Further, improved structural integrity of outputs by using a pre-training scheme.

Course / Independent Projects
project image Advanced Computer Graphics (CS-440)
Report / Slides

Extended the barebones raytracer, Nori for the Advanced Computer Graphics (CS-440) course with features including homogenous participating media and won the runners-up prize in the rendering competition.

project image Learn.ai
Code

Webapp to constuct ML models or neural networks in a "no-code" fashion, by dragging and dropping layers.


Blogs
project image Graph Convolutional Networks: A Theoretical Deep Dive
Blog

Blogs delves into the structure, underlying principles, and forward propagation rule of graph convolutional networks.


Design and source code from Jon Barron's website