Research Interests & Activities

Debjit Daw

Xilinx ZCU104 Development

Xilinx ZCU104 Development

Xilinx ZCU104 Development

Medical Image seg

Medical Image seg

Medical Image seg

Japan Exchange Program

Japan Exchange Program

Japan Exchange Program

Edge AI Testing

Edge AI Testing

Edge AI Testing

Team Collaboration

Team Collaboration

Team Collaboration

Conference Presentation

Conference Presentation

Conference Presentation

Model Training

Model Training

Model Training

Overview

Mathematics and Biological sciences always fascinated me, but computing was an acquired taste, which is why I studied Information Technology (IT) in my Bachelors degree. It was not only limited to computer architecture, but also extended to statistical applications for information processing. During this time, With sheer curiosity, I explored Android and web development, self- learned C# for game development, and tinkered around with Raspberry Pi, deploying a home server during vacations.I have also been working on a personal project on Xilinx Vitis AI runtime, where I am focused on quantising vision models like YOLOv8 and deploying them on FPGA boards like the ZCU104.

Research Topics

Computer Vision Applications

I am really interested in the domain of computer vision and constantly exploring this, by working closely with my professors

Ongoing

2 Publications

Vision TransformersAction RecognitionConvolution Neural NetworksRoad Safety
Related ProjectsPublications

Large Language Models Orchestration

I have worked with LLMs for Nutrional Analysis and currently exploring further in this field.

Ongoing

1 Publication

Multi-Agent SystemsKnowledge GraphsOCRLLM Integration
Related ProjectsPublications

IoT & Smart Infrastructure

I am focussed on researching on IoT devices, to their maximum capacity and have worked with sensor and computer vision based tasks for road infrastructure management and driver safety

2 Publications

IoTSmart CitiesSensor NetworksEdge ProcessingADAS
Related ProjectsPublications

Vision Model Quantization & Edge AI Deployment

Developing efficient vision models for edge devices via quantization techniques, ONNX Runtime optimization as well as VITIS AI based quantization for deployment on platforms like Xilinx ZCU104.

Ongoing

0 Publications

Model QuantizationONNX RuntimeEdge ComputingHardware Acceleration
Related ProjectsPublications

Healthcare Assistance Systems

I have worked with medical images and developed a Cloud-based diagnostic systems for overian cancer detection. I am really egar to contribute further in this domain

Ongoing

0 Publications

Medical ImagingCloud InfrastructureDiagnostic SystemsDeep Learning
Related ProjectsPublications