Wipro Computer Vision Research
Solved open computer vision problems across floor-plan recognition, COVID audio detection, and document denoising for enterprise and government clients.
Pranav Pandey
Senior Applied AI Engineer at SpotDraft
From their time as
Senior Research Engineer, Machine Learning and Computer Vision
Wipro • 2020 - 2021
Overview
At Wipro, Pranav worked on a range of open computer vision problems across different modalities and client contexts. For a government authority in Dubai, he built a module for automatic floor-plan understanding and recognition of objects of interest, designed to fast-track the plan approval process.
The Story
At Wipro, Pranav worked on a range of open computer vision problems across different modalities and client contexts. For a government authority in Dubai, he built a module for automatic floor-plan understanding and recognition of objects of interest, designed to fast-track the plan approval process.
In parallel, he contributed to a COVID detection system that identified COVID-positive patients from cough audio recordings, working across a team effort that combined audio signal processing with deep learning classification. He also developed a document denoising system using a modification of CycleGAN, designed to generalize across different types of dirty or aged documents.
Additionally, he worked on zero-shot super resolution, where a super-resolved image could be generated for any input image by training on that image alone, without requiring a higher-resolution reference. This work required understanding and adapting research-level techniques for practical application.