promnt.ai.txt

Help me write the code for loop trough the images in the folder and apply the SAHI prediction on each image.
The output should be the predicted bounding box for each image save in json format. And export the prediction result to a csv file.
then print out the object detection count for each class for each image and totaled for all images.

As some of the objects are circle or rectangle, can we use the cv2 utils or pixel checking to refine the bonunding box prediction? Such as scaning the pixel image and move the bounding box to cover the outline of rectangle inside the bondung box.