test setup

yolov7-setup/vott2yolov.py

@@ -0,0 +1,75 @@
+import numpy as np
+import json, sys, os, shutil
+
+## cliquer sur export dans Vott puis :
+
+# prefix = "../exports_VOTT/CH01-20230614-083436-091514-Zone2/vott-json-export/"
+# vottjson_filename = "Poulets-export.json"
+# cc = 1000000
+
+prefix = "/Users/noham/Documents/GitHub/Stage-2024/yolov7-tracker/"
+print('Working folder: ', prefix)
+jsonname = 'vott-json-export'
+jsonfile = 'track-2fps-export.json'
+json_path = prefix + jsonname + '/' + jsonfile
+print('Working json file: ', json_path)
+jsonfolder = prefix+jsonname
+print('Working json folder: ', jsonfolder)
+
+outputname = 'dataset2fps'
+print('Ooutput folder: ', outputname)
+output = '/Users/noham/Documents/GitHub/Stage-2024/test/yolov7-tracker/data/'+outputname
+print('Output folder: ', output)
+cc = 0
+
+label = 'runner'
+
+for folder in [output, output+"/images",output+"/labels"]:
+    if not os.path.exists(folder):
+        os.makedirs(folder)
+
+with open(json_path, "r") as read_file:
+    data = json.load(read_file)
+
+imglist = []
+for a in data["assets"]:
+    asset = data["assets"][a]["asset"]
+    regions = data["assets"][a]["regions"]
+    img_name = asset["name"]
+    width = asset["size"]["width"]
+    height = asset["size"]["height"]
+    # print("\nimage name : ",img_name," width =",width," height =",height)
+    shutil.copyfile(jsonfolder+"/"+img_name,output+"/images/image_"+str(cc).zfill(12)+".jpg")
+    imglist.append(output+"/images/image_"+str(cc).zfill(12)+".jpg")
+    f = open(output+"/labels/image_"+str(cc).zfill(12)+".txt", "w")
+    # f = open(prefix+"/"+img_name.replace(".jpg",".txt"), "w")
+    for region in regions:
+        boundingBox = region["boundingBox"]
+        points = region["points"]
+        # print("    tags =",region["tags"]," boundingBox =",boundingBox," points =",points)
+        # ## Labels :
+        # D’après la doc on doit avoir :
+        # Label_ID_1 X_CENTER_NORM Y_CENTER_NORM WIDTH_NORM HEIGHT_NORM
+        # The label_id is the index number in the classes.names file. The id of the first label will be 0 and an increasing integer after that. Note all the position attributes in the label file are not absolute but normalised.
+        # X_CENTER_NORM = X_CENTER_ABS/IMAGE_WIDTH   centre image
+        # Y_CENTER_NORM = Y_CENTER_ABS/IMAGE_HEIGHT  centre image
+        # WIDTH_NORM = WIDTH_OF_LABEL_ABS/IMAGE_WIDTH    largeur boite
+        # HEIGHT_NORM = HEIGHT_OF_LABEL_ABS/IMAGE_HEIGHT hauteur boite
+        # dans un repere dont l'origine est en haut a gauche et X de gauche à droite, Y du haut vers le bas
+        Label_ID = "0"
+        X_CENTER_NORM = (boundingBox["left"]+boundingBox["width"]/2)/width
+        Y_CENTER_NORM = (boundingBox["top"]+boundingBox["height"]/2)/height
+        WIDTH_NORM = boundingBox["width"]/width
+        HEIGHT_NORM = boundingBox["height"]/height
+        f.write(Label_ID+" "+str(X_CENTER_NORM)+" "+str(Y_CENTER_NORM)+" "+str(WIDTH_NORM)+" "+str(HEIGHT_NORM)+"\n" ) # 0 = poulet si l'on a qu'une classe d'objet...
+    f.close()
+    cc += 1
+    # print("    written file :",prefix+"/"+img_name.replace(".jpg",".txt"))
+with open(output+"/liste_images.txt", "w") as file:
+    for item in imglist:
+        file.write(item + "\n")
+with open(output+"/dataset.yaml", "w") as file:
+    file.write(f"""train: data/{outputname}/liste_images.txt
+val: data/{outputname}/liste_images.txt
+nc: 1
+names: ['{label}']""")