comments measure mean average precision

measure_map.py

@@ -76,11 +76,12 @@ def get_map(pred, gt, f):
 				break
 			else:
 				continue
-
+		print(found_match)
 		T[pred_class].append(int(found_match))
 
 	for gt_box in gt:
 		#if not gt_box['bbox_matched'] and not gt_box['difficult']:
+		# In case the GT was not found by the network
 		if not gt_box['bbox_matched']:
 			if gt_box['class'] not in P:
 				P[gt_box['class']] = []
@@ -135,7 +136,7 @@ C.model_path = 'data/model_frcnn.hdf5'
 
 img_path = options.test_path
 
-
+# Adapt size of image to input size of network and return also the resizing values
 def format_img(img, C):
 	img_min_side = float(C.im_size)
 	(height,width,_) = img.shape
@@ -209,8 +210,8 @@ all_imgs, _, _ = get_data(options.test_path)
 test_imgs = all_imgs
 
 
-T = {}
-P = {}
+T = {} # Truth
+P = {} # Prediction
 for idx, img_data in enumerate(test_imgs):
 	print('{}/{}'.format(idx,len(test_imgs)))
 	st = time.time()
@@ -218,6 +219,9 @@ for idx, img_data in enumerate(test_imgs):
 
 	img = cv2.imread(filepath, cv2.IMREAD_IGNORE_ORIENTATION | cv2.IMREAD_COLOR)
 
+	# X is the resized image
+	# fx is the resizing factor along the x axis
+	# fy is the resizing factor along the y axis
 	X, fx, fy = format_img(img, C)
 
 	if K.image_dim_ordering() == 'tf':