Ankit Kumar

Senior Software Engineer-R&D

TITAN COMPANY LIMITED•  August 2023 - Present•  Bangalore, India

Implemented YOLO v8 model to segment diamonds in studded jewellery, achieving an accuracy of 95%. Utilized Mask-RCNN with ResNet-101 backbone and U-Net models to detect and mask sparks emitted from diamonds and developed a system to quantify sparks using advanced computer vision techniques. Applied YOLO v7-X model for counting diamonds across various shapes, sizes, and surface types, achieving an accuracy of 98%. Built a machine to analyze the light performance of diamonds, leveraging machine vision and advanced computer vision techniques. Created a sorting system using computer vision and signature mapping algorithms to detect the shape and size of fancy diamonds, achieving an accuracy of 98%.

Deep Learning Intern

SYNERGY LABS•  November 2022 - January 2023•  Gurugram, India

Developed and deployed an Automatic Traffic Counter and Classifier (ATCC) using Deepstream SDK and YOLOv4-Tiny models, achieving a remarkable 92% accuracy in vehicle classification and counting. Significantly improved the Automatic Number Plate Detection System, increasing accuracy by up to 90% through script enhancements. Optimized system performance by deploying models on Jetson Xavier NX, establishing a potent edge system for efficient traffic analysis.

Machine Learning Intern

APESAI•  April 2022 - July 2022•  Remote

Implemented a robust smoke and fire detection system from industrial CCTV using Computer Vision and EfficientNet V2 model, achieving an 88% accuracy. Explored NLP and TIME-SERIES, applying ARIMAX and SARIMAX for stock market prediction with 85% accuracy through textual analysis. Developed an OpenCV-based Social Distancing Detector using object detection and Euclidean distancing techniques.