formatting

box_converter.py

@@ -4,21 +4,22 @@ import pycocotools.mask
 
 
 def x1y1x2y2_2_x1x2y1y2(this_list):
-	new_list = [0]*4
-	new_list[0] = this_list[0]
-	new_list[1] = this_list[2]
-	new_list[2] = this_list[1]
-	new_list[3] = this_list[3]
-	return new_list
+    new_list = [0] * 4
+    new_list[0] = this_list[0]
+    new_list[1] = this_list[2]
+    new_list[2] = this_list[1]
+    new_list[3] = this_list[3]
+    return new_list
 
 
 def x1x2y1y2_2_x1y1x2y2(this_list):
-	new_list = [0]*4
-	new_list[0] = this_list[0]
-	new_list[1] = this_list[2]
-	new_list[2] = this_list[1]
-	new_list[3] = this_list[3]
-	return new_list
+    new_list = [0] * 4
+    new_list[0] = this_list[0]
+    new_list[1] = this_list[2]
+    new_list[2] = this_list[1]
+    new_list[3] = this_list[3]
+    return new_list
+
 
 def xywh2xyxy_yolo(x):
     y = np.copy(x)
@@ -29,15 +30,15 @@ def xywh2xyxy_yolo(x):
     return y
 
 
-
 def xywh2xxyy(list_input):
     this_list = list_input.copy()
-    this_list[0] = list_input[0]					# x1
-    this_list[1] = list_input[0] + list_input[2]	# x2
-    this_list[2] = list_input[1]					# y1
-    this_list[3] = list_input[1] + list_input[3]	# y2
+    this_list[0] = list_input[0]  # x1
+    this_list[1] = list_input[0] + list_input[2]  # x2
+    this_list[2] = list_input[1]  # y1
+    this_list[3] = list_input[1] + list_input[3]  # y2
     return this_list
 
+
 def xywh2xyxy(list_input):
     this_list = list_input.copy()
     this_list[2] = list_input[0] + list_input[2]
@@ -46,56 +47,74 @@ def xywh2xyxy(list_input):
     this_list[1] = list_input[1]
     return this_list
 
-def xyxy2xywh(list_input):
-	x1 = int(list_input[0])
-	y1 = int(list_input[1])
-	w = abs(int(list_input[2]-list_input[0]))
-	h = abs(int(list_input[3]-list_input[1]))
-	return [x1,y1,w,h]
+
+def xyxy2xywh(listinput):
+    x1 = int(listinput[0])
+    y1 = int(listinput[0])
+    w = int(listinput[2] - listinput[0])
+    h = int(listinput[3] - listinput[1])
+    assert w > 0
+    assert h > 0
+    return [x1, y1, w, h]
 
 
 def xywh_2_x1y1x2y2(list_input):
-	x1 = int(list_input[0])
-	y1 = int(list_input[1])
-	x2 = int(x1 + list_input[2])
-	y2 = int(y1 + list_input[3])
-	return [x1,y1,x2,y2]
-
-
-def x1y1x2y2_2_cxcywh(list_input,height_img, width_img):
-	# conversin of [x1,y1,x2,y2] to normalized yolo! [center_x,center_y,width,height]
-	center_x = (list_input[0]+list_input[2])/2
-	center_y = (list_input[1]+list_input[3])/2
-	width = (list_input[2]-list_input[0])
-	height = (list_input[3]-list_input[1])
-	return [center_x/width_img,center_y/height_img,width/width_img,height/height_img]
-
-def x1y1wh_2_cxcywh(list_input,height_img, width_img):
-	# conversin of [x1,y1,x2,y2] to normalized yolo! [center_x,center_y,width,height]
-	center_x = (list_input[0]+list_input[2]*0.5)
-	center_y = (list_input[1]+list_input[3]*0.5)
-	width = list_input[2]
-	height = list_input[3]
-	return [center_x/width_img,center_y/height_img,width/width_img,height/height_img]
+    x1 = int(list_input[0])
+    y1 = int(list_input[1])
+    x2 = int(x1 + list_input[2])
+    y2 = int(y1 + list_input[3])
+    return [x1, y1, x2, y2]
+
+
+def x1y1x2y2_2_cxcywh(list_input, height_img, width_img):
+    # conversin of [x1,y1,x2,y2] to normalized yolo! [center_x,center_y,width,height]
+    center_x = (list_input[0] + list_input[2]) / 2
+    center_y = (list_input[1] + list_input[3]) / 2
+    width = list_input[2] - list_input[0]
+    height = list_input[3] - list_input[1]
+    return [
+        center_x / width_img,
+        center_y / height_img,
+        width / width_img,
+        height / height_img,
+    ]
+
+
+def x1y1wh_2_cxcywh(list_input, height_img, width_img):
+    # conversin of [x1,y1,x2,y2] to normalized yolo! [center_x,center_y,width,height]
+    center_x = list_input[0] + list_input[2] * 0.5
+    center_y = list_input[1] + list_input[3] * 0.5
+    width = list_input[2]
+    height = list_input[3]
+    return [
+        center_x / width_img,
+        center_y / height_img,
+        width / width_img,
+        height / height_img,
+    ]
+
 
 def binary_mask_to_rle(binary_mask):
-	## input is numpy array with either 0 or 1 : np.array([0,0,1,1,1,0,1])
-	## output: rle={'counts': [2, 3, 1, 1], 'size': [7]}
-    rle = {'counts': [], 'size': list(binary_mask.shape)}
-    counts = rle.get('counts')
-    for i, (value, elements) in enumerate(groupby(binary_mask.ravel(order='F'))):
+    ## input is numpy array with either 0 or 1 : np.array([0,0,1,1,1,0,1])
+    ## output: rle={'counts': [2, 3, 1, 1], 'size': [7]}
+    rle = {"counts": [], "size": list(binary_mask.shape)}
+    counts = rle.get("counts")
+    for i, (value, elements) in enumerate(groupby(binary_mask.ravel(order="F"))):
         if i == 0 and value == 1:
             counts.append(0)
         counts.append(len(list(elements)))
     return rle
 
+
 def rle_to_binary_mask(rle):
-	## converts rle={'counts': [2, 3, 1, 1], 'size': [7]} to np.array([0,0,1,1,1,0,1])
-	compressed_rle = pycocotools.mask.frPyObjects(rle, rle.get('size')[0], rle.get('size')[1])
-	mask = pycocotools.mask.decode(compressed_rle)
-	return mask
+    ## converts rle={'counts': [2, 3, 1, 1], 'size': [7]} to np.array([0,0,1,1,1,0,1])
+    compressed_rle = pycocotools.mask.frPyObjects(
+        rle, rle.get("size")[0], rle.get("size")[1]
+    )
+    mask = pycocotools.mask.decode(compressed_rle)
+    return mask
 
 
 def box2centroid(b):
-    centroid = [int((b[2]+b[0])/2), int((b[3]+b[1])/2)]
+    centroid = [int((b[2] + b[0]) / 2), int((b[3] + b[1]) / 2)]
     return centroid