Projects Implemented for the Udacity Sensor Fusion Engineer Nanodegree Program

Build the feature tracking part of a collision detection system, and test various combinations of keypoint detectors and descriptors to see which combinations perform best.

A high-performance, multi-threaded C++ pipeline for real-time multi-camera keypoint detection.

Final Project of Computer Vision Exam (A.Y. 2022/2023) from Prof. Zanuttigh

Keypoint-matching (AKAZE method) using OpenCV library

Feature tracking using keypoints and descriptors

Minimal-dependency Computer Vision & Image Processing library built in C++17.

Testing various detector / descriptor combinations to see which ones perform best to be used in a collision detection system. Also 2 different approaches (FLANN vs. Brute-force with the descriptor distance ratio test) for keypoints matching are tested.

ZED SDK Example projects

Tracking the preceding vehicle using Lidar and camera sensors to calculate the Time To Collision (TTC).

Compute time-to-collision (TTC) using Lidar and Camera sensors. Identify suitable keypoint detector-descriptor combinations for TTC estimation.

Face Morphing Java Android app.

WACV 2023: Centroid Distance Keypoint Detector for Colored Point Clouds

使用ONNXRuntime部署DeDoDe："局部特征匹配：检测，不要描述——描述，不要检测"。依然是C++和Python两个版本的程序

使用retinaface完成车牌关键点检测，并在tensorRT下部署

OpenPose: Real-time multi-person keypoint detection library for body, face, hands, and foot estimation