The whole process of computer vision learning

learning schedule

1 preliminary understanding

• Test practice : YOLOV5 environment construction and training COCO128 data set , YOLO V7 training own data set
• Tracking practice : YOLOV5+DeepSort multi-target tracking (pedestrians)
• Speed ​​measurement practice : Multi-vehicle tracking and speed measurement based on YOLOv5 + Deepsort

2 Theoretical knowledge

2.1 Object Detection

2.1.1 BP neural network

• BP neural network (Python implementation)

• BP algorithm of neural network (illustration + theoretical derivation + example application + code)

• convolutional neural network (CNN)

2.1.2 Traditional Machine Vision

• SVM, decision tree, random forest, adboost

Deep Learning Machine Vision

• Feature extraction network (alxnet + vggnet + densenet, etc.)
• Target detection algorithm (R-CNN, fast R-CNN, faster R-CNN, yolo, SSD, yoloV2, yoloV3, yoloV4, yoloV5, yoloV6, yoloV7)

2.2 Target Tracking

• DeepSORT multi-target tracking principle
• Kalman filter + Hungarian matching

2.3 Camera Calibration

• Transformation of the four major coordinate systems and camera calibration (theory plus practical operation of the chessboard method)

2.4 Speed ​​measurement

Extracting Vessel Speed Based on Machine Learning and Drone Images during Ship Traffic Flow Prediction