Author:Ziyan Jiang([email protected]), Ying Zhu, Jiayi Jin
Yolov5(You Only Look Once Version 5) is an advanced object detection model, . YOLOv5 builds on the YOLO (You Only Look Once) series of models and is capable of performing object detection tasks in a single forward pass, simultaneously predicting multiple objects in an image along with their bounding boxes and classes.
Here are the steps to run the task.
First, we resize the images that we take to 640*640, to make it sutiable for the following task, you could see the code in script resize_images.py. You could use it by just changing the data location.
To change the origin images to annotations, we use the tool LabelImg to mark the data, the code is in the folder labelimg.
We use yolov5 to train the model, the image in folder yolov5/course_design/images, the labels in folder yolov5/course_design/labels, you can change training set, validation set, number of classes, class names in info.yaml.
To train the model, adjust the config in run_course_design.sh, and input following code in terminal:
sh run_course_design.shAfter training, you will get the result in folderyolov5/runs/train.