I'm developing an AI-powered pipeline for automating construction site management. This included object detection and activity recognition with the creation of Construction Action dataset to automate the generation of Construction Daily Reports (CDRs), reducing manual effort and improving safety compliance monitoring on-site.

● Fish-Eye Camera-Based Vehicle Detection and License Plate RecognitionDeveloped a C++ application for vehicle detection and license plate recognition using fisheye cameras. Created image stitching algorithms and addressed challenges like extreme weather. Awarded Best Team and Best Effort for driving the highest revenue for six months.● Human Action Recognition for Fisheye CamerasWorked on human action recognition with a small dataset by using fisheye data augmentation and realistic human models from game engines to enhance performance.

● OCR and Steel Pipe Counting ApplicationsImplemented a deep learning model for steel pipe counting, reducing inspection time by 20% through efficient dataset creation and annotation.Developed an OCR system to read embossed text on metal surfaces, optimizing processing time with computer vision techniques and integrating Google Vision API for text recognition.