Publications
DPPD: Deformable Polar Polygon Object Detection
Deformable Polar Polygon Object Detection (DPPD) predicts sparse, flexible vertices in polar coordinates for precise, efficient object detection without NMS, enabling real-time autonomous driving applications.
Unlimited Road-scene Synthetic Annotation (URSA) Dataset
Synthetic semantic segmentation dataset containing upwards of a million images generated with a method capable of automatically producing unique, on-demand data annotations in real-time.
A Hierarchical Deep Architecture and Mini-Batch Selection Method For Joint Traffic Sign and Light Detection
Deep hierarchical architecture in conjunction with a mini-batch proposal selection mechanism that allows a network to detect both traffic lights and signs from training on separate traffic light and sign datasets.
Projects
Capstone Design Project
Helped build a learning-based autonomous lane keeping system for a final year engineering design project. The end result was deployed onto a modified Toyota Corolla, allowing the car to autonomously steer on various urban roads in the Kitchener-Waterloo region with only a single monocular camera.
Fast Semantic Segmentation
Full Tensorflow implementation of the work described in ”ICNet for Real-Time Semantic Segmentation on High-Resolution Images”. Only project to successfully implement the training and compression process.
Face Block
Won first place in Qualcomm’s Intern Hackathon with an Android app that uses a custom trained MobileNet SSD detector and MIL tracker to block out faces in videos in real-time and on-device.
Realty Sheets
Realty Sheets is a cloud based software tool that helps Real Estate agents manage their Open Houses using customizable sign-in sheets.