dikshap07/vehicle_and_person_detection

YOLO project

Model darknet yolov3 416

directories:

1. Code:

   project/vehicle_and_person_detection/code
   
   
   for image :
               file: yolo_image.py
   
               command: python yolo_image.py 
   
               for save the detected image: python yolo_image.py --save True
              
               Dependent folder: code/input:- handlabel(40 images) , coco(validation2017) specific to vehicle and person
   
               Takes images from handlabel_images or coco_validation_images 
               save detected bounding box text files handlabel_detected-results or coco_detected-results
               save detected images detected-images
   
   for video : 
               file: yolo_video.py
   
               command: python yolo_output.py --input <file-name> --output 
   
               for save the text (detected bounding) file separately python yolo_output.py --input <file-name> --output --save_file_separately
   
               for save the text (detected bounding) file python yolo_output.py --input <file-name> --output --save_file
   
               video save in out_video folder
               input video path : video (todo) presently code
   
   for mAP :
              file: calculate_map.py
             
              command: python calculate_map.py
   
              folder requirement: coco_ground-truth, coco_detected-results , coco_validation_images(optional)
              result: results/<filename>
   

train :

 file : train.py
 
 command: pthon train.py

 Note: change weights, anchor and classes path in train.py as per the need