Autonomous Vehicle Image & Video Annotation Project
Led a large-scale computer vision data annotation project supporting ADAS model training for an autonomous mobility client. The project involved high-precision bounding box annotation, semantic and instance segmentation of road elements, lane markings, pedestrians, vehicles, traffic signs, and object tracking across video sequences. The dataset included over 250,000 images and 3,000+ video sequences captured across diverse environmental conditions (day/night, rain, low-light, urban and highway scenarios). Implemented a multi-layer quality assurance workflow including peer review, senior QA validation, and periodic accuracy audits to maintain >98% annotation accuracy. The team operated in structured day and night shifts to ensure continuous delivery and faster turnaround times while adhering to strict data security and NDA compliance protocols.