Vardhan Mistry

Focus: Camera Soiling Detection & Robustness | Grade: 1.0 • Developed a real-time semantic segmentation framework for ADAS camera soiling detection. • Engineered MudGAN to bridge the sim-to-real gap via synthetic data generation. • Benchmarked CNN and Vision Transformer (ViT) backbones for accuracy-latency trade-offs. Achieved 0.94 Dice and 0.89 IoU, reducing False Positives by 95%. • Built a evaluation pipeline using FID, KID, and LPIPS metrics. • Optimized inference to 7 ms on NVIDIA GPUs using ONNX.

Focus: Synthetic Rain Augmentation • Designed a physics-inspired procedural rain generation pipeline for robust data augmentation. • Modeled synthetic camera blindness scenarios to simulate and evaluate perception safety. • Conducted controlled experiments to assess detection robustness in adverse weather conditions. • Delivered actionable insights for algorithm improvement through detailed performance analysis reports.

Focus: Automation & Robotics Systems • Led the deployment of large-scale automation systems in a greenfield industrial project. • Integrated and commissioned a FANUC industrial robotic system. • Coordinated cross-functional teams to deliver complex system-level milestones. • Developed Power BI dashboards for data-driven operational tracking and executive reporting.

Focus: Computer Vision, Robotics, Automation, ADAS

Majors: • Computer Application in Mechanical Engineering • Computer Techniques and Programming Language • Control System • Electrical and Electronics • Industrial Engineering