kennychen880127 Goto Github PK

fatigue-driving-detection-using-lstm-of-facial-landmarks

This project implements a system that utilizes dlib for facial feature extraction to calculate Eye Aspect Ratio (EAR). Subsequently, it employs LSTM for predicting whether the driver exhibits signs of fatigue based on facial features.

implementation-of-a-building-advertising-using-instance-segmentation-and-augmented-reality

This experiment utilizes YOLACT for road scene recognition. It employs two ZED 2i stereo cameras for 3D point cloud calculations, enabling a forward safety warning system by determining object-to-vehicle distances. Additionally, it identifies departmental buildings for augmented reality advertisements.

kennychen880127

labelimg-only-for-yolo

Modifying and simplifying the functionality of LabelImg, this version exclusively supports the YOLO annotation format. It features automatic saving, auto-configuration of the annotation storage format, and validation checks to ensure the input class matches the default value.

lane-detection-system-with-augmented-reality-alerts

Implementing road detection through OpenCV, and integrating augmented reality warning methods to provide drivers with more visual alerts.

mlp-corrected-distance-estimation-for-non-horizontal-camera-measurements

When the camera and the target object are not horizontally aligned for distance measurement, the Multilayer Perceptron (MLP) is utilized. It takes screen coordinates and the current measured distance as inputs to correct and predict the horizontal distance of the camera.

monitoring-gui-states-with-yolov8-object-detection

This project utilizes YOLOv8 for GUI monitoring, enabling the surveillance of multiple remote machine screen values and statuses.

recording-with-multiple-cameras

This project enables UI-controlled addition, editing, and deletion of cameras, and it utilizes threads and FrameBuffer to simultaneously open multiple cameras.

yolov8-improved-pose-estimation

This project utilizes the Intel RealSense D435i camera to measure the distance of keypoints in YOLOv8's pose estimation. It employs the Quartile algorithm to identify outliers in the distance data and compensates for keypoints and distances accordingly.

yolov8-py-to-exe

Use Pyinstaller to package the YOLOv8 project into an executable (.exe) file.

yolov8-tank-tracking

Using the YOLOv8 tracking module to identify tank IDs and achieve tracking by selecting IDs with a simple user interface integrated with mouse input.