(computer vision) IOU 구하기

## IOU 구하기
## 입력인자로 후보 박스와 실제 박스를 받아서 IOU를 계산하는 함수 생성

import numpy as np 

def compute_iou(cand_box, gt_box):

    # Calculate intersection areas
    x1 = np.maximum(cand_box[0], gt_box[0])
    y1 = np.maximum(cand_box[1], gt_box[1])
    x2 = np.minimum(cand_box[2], gt_box[2])
    y2 = np.minimum(cand_box[3], gt_box[3])
    
    intersection = np.maximum(x2 - x1, 0) * np.maximum(y2 - y1, 0)
    
    cand_box_area = (cand_box[2] - cand_box[0]) * (cand_box[3] - cand_box[1])
    gt_box_area = (gt_box[2] - gt_box[0]) * (gt_box[3] - gt_box[1])
    union = cand_box_area + gt_box_area - intersection
    
    iou = intersection / union
    return iou

import cv2
import matplotlib.pyplot as plt
%matplotlib inline

# 실제 box(Ground Truth)의 좌표를 아래와 같다고 가정. 
gt_box = [60, 15, 320, 420]

# 코랩 버전은 상대 경로를 사용하지 않습니다. /content 디렉토리를 기준으로 절대 경로를 이용합니다. 
# default_dir 은 /content/DLCV로 지정하고 os.path.join()으로 상세 파일/디렉토리를 지정합니다. 
default_dir = '/content/DLCV'
img = cv2.imread('./data/audrey01.jpg')
img_rgb = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)

red = (255, 0 , 0)
img_rgb = cv2.rectangle(img_rgb, (gt_box[0], gt_box[1]), (gt_box[2], gt_box[3]), color=red, thickness=2)

plt.figure(figsize=(8, 8))
plt.imshow(img_rgb)
plt.show()

 

import selectivesearch

#selectivesearch.selective_search()는 이미지의 Region Proposal정보를 반환 
_, regions = selectivesearch.selective_search(img_rgb, scale=100, min_size=2000)

print(type(regions), len(regions))

cand_rects = [cand['rect'] for cand in regions]
for index, cand_box in enumerate(cand_rects):
    cand_box = list(cand_box)
    cand_box[2] += cand_box[0]
    cand_box[3] += cand_box[1]
    
    iou = compute_iou(cand_box, gt_box)
    print('index:', index, "iou:", iou)
    
    index: 0 iou: 0.0
index: 1 iou: 0.0713319505816575
index: 2 iou: 0.1134453781512605
index: 3 iou: 0.9874899187876287
index: 4 iou: 0.9748907882241216
index: 5 iou: 0.09851851851851852
index: 6 iou: 0.04228869895536562
index: 7 iou: 0.15042735042735042
index: 8 iou: 0.14541310541310543
index: 9 iou: 0.10112060778727446
index: 10 iou: 0.1194681861348528
index: 11 iou: 0.14017094017094017
index: 12 iou: 0.0
index: 13 iou: 0.0
index: 14 iou: 0.5933903133903133
index: 15 iou: 0.0
index: 16 iou: 0.0853656220322887
index: 17 iou: 0.12535612535612536
index: 18 iou: 0.0
index: 19 iou: 0.1498762283399595
index: 20 iou: 0.0
index: 21 iou: 0.05242165242165242
index: 22 iou: 0.0
index: 23 iou: 0.050123456790123457
index: 24 iou: 0.07313390313390314
index: 25 iou: 0.009348810110697799
index: 26 iou: 0.0
index: 27 iou: 0.23817663817663817
index: 28 iou: 0.18719489628676664
index: 29 iou: 0.19555555555555557
index: 30 iou: 0.20454890788224123
index: 31 iou: 0.0
index: 32 iou: 0.5958024691358025
index: 33 iou: 0.0
index: 34 iou: 0.1515749506657008
index: 35 iou: 0.5958024691358025
index: 36 iou: 0.1134453781512605
index: 37 iou: 0.354069104098905
index: 38 iou: 0.0
index: 39 iou: 0.1134453781512605
index: 40 iou: 0.12107738349722104
index: 41 iou: 0.3278419532685744
index: 42 iou: 0.3837088388214905
index: 43 iou: 0.2757834757834758
index: 44 iou: 0.3956795484151107
index: 45 iou: 0.12107738349722104
index: 46 iou: 0.2113350294590252
index: 47 iou: 0.5008648690956052
index: 48 iou: 0.7389566501483806
index: 49 iou: 0.815085997397344
index: 50 iou: 0.6270619201314865
index: 51 iou: 0.6270619201314865
index: 52 iou: 0.6270619201314865

cand_rects = [cand['rect'] for cand in regions if cand['size'] > 5000]
cand_rects.sort()
cand_rects

[(0, 0, 58, 257),
 (0, 0, 58, 266),
 (0, 0, 58, 266),
 (0, 0, 171, 449),
 (0, 0, 195, 449),
 (0, 0, 374, 449),
 (0, 0, 374, 449),
 (0, 0, 374, 449),
 (0, 256, 57, 188),
 (0, 256, 171, 193),
 (16, 0, 270, 50),
 (16, 0, 270, 109),
 (16, 0, 270, 199),
 (16, 0, 302, 418),
 (62, 17, 256, 401),
 (63, 0, 311, 449),
 (63, 18, 114, 91),
 (63, 91, 120, 123),
 (63, 91, 132, 220),
 (63, 171, 254, 246),
 (63, 171, 254, 247),
 (63, 171, 254, 247),
 (63, 171, 311, 278),
 (84, 313, 290, 136),
 (84, 360, 290, 89),
 (84, 360, 290, 89),
 (134, 18, 119, 181),
 (136, 23, 117, 176),
 (153, 366, 151, 51),
 (166, 23, 87, 176),
 (197, 0, 177, 420),
 (197, 18, 120, 132),
 (197, 18, 120, 209),
 (232, 79, 85, 148),
 (284, 0, 90, 420),
 (284, 0, 90, 420),
 (284, 0, 90, 420),
 (323, 105, 51, 248)]

img = cv2.imread('./data/audrey01.jpg')
img_rgb = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)
print('img shape:', img.shape)

green_rgb = (125, 255, 51)
cand_rects = [cand['rect'] for cand in regions if cand['size'] > 3000]
gt_box = [60, 15, 320, 420]
img_rgb = cv2.rectangle(img_rgb, (gt_box[0], gt_box[1]), (gt_box[2], gt_box[3]), color=red, thickness=2)

for index, cand_box in enumerate(cand_rects):
    
    cand_box = list(cand_box)
    cand_box[2] += cand_box[0]
    cand_box[3] += cand_box[1]
    
    iou = compute_iou(cand_box, gt_box)
    
    if iou > 0.5:
        print('index:', index, "iou:", iou, 'rectangle:',(cand_box[0], cand_box[1], cand_box[2], cand_box[3]) )
        cv2.rectangle(img_rgb, (cand_box[0], cand_box[1]), (cand_box[2], cand_box[3]), color=green_rgb, thickness=1)
        text = "{}: {:.2f}".format(index, iou)
        cv2.putText(img_rgb, text, (cand_box[0]+ 100, cand_box[1]+10), cv2.FONT_HERSHEY_SIMPLEX, 0.4, color=green_rgb, thickness=1)
    
plt.figure(figsize=(12, 12))
plt.imshow(img_rgb)
plt.show()

img shape: (450, 375, 3)
index: 3 iou: 0.9874899187876287 rectangle: (59, 14, 321, 421)
index: 4 iou: 0.9748907882241216 rectangle: (62, 17, 318, 418)
index: 12 iou: 0.5933903133903133 rectangle: (63, 171, 317, 417)
index: 27 iou: 0.5958024691358025 rectangle: (63, 171, 317, 418)
index: 30 iou: 0.5958024691358025 rectangle: (63, 171, 317, 418)
index: 42 iou: 0.5008648690956052 rectangle: (63, 171, 374, 449)
index: 43 iou: 0.7389566501483806 rectangle: (63, 0, 374, 449)
index: 44 iou: 0.815085997397344 rectangle: (16, 0, 318, 418)
index: 45 iou: 0.6270619201314865 rectangle: (0, 0, 374, 449)
index: 46 iou: 0.6270619201314865 rectangle: (0, 0, 374, 449)
index: 47 iou: 0.6270619201314865 rectangle: (0, 0, 374, 449)