BPPC/visualize_bppc.py at main · OSHMOS/BPPC · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
import os
import re
import cv2
import copy
import torch
import numpy as np
from math import sqrt

torch.cuda.empty_cache()

class Visualize_BPPC():
    # def __init__(self, handed_option, bppc_kpts, sm_kpts, gt_kpts, folder_number):
    def __init__(self, handed_option, bppc_kpts, hrnet_kpts, sm_kpts, gt_kpts, folder_number):
        super().__init__()
        self.handed_option = handed_option

        self.bppc_kpts = bppc_kpts
        self.hrnet_kpts = hrnet_kpts
        self.sm_kpts = sm_kpts
        self.gt_kpts = gt_kpts

        self.folder_number = folder_number


    def sort_key(self, s):
        # 이미지 파일명에서 숫자를 추출하여 정렬 기준으로 사용
        return int(re.search(r'\d+', s).group())


    def show2Dpose(self, kps, img):
        connections = [[0, 1], [1, 2], [2, 3], [0, 4], [4, 5],
                    [5, 6], [0, 7], [7, 8], [8, 9], [9, 10],
                    [8, 11], [11, 12], [12, 13], [8, 14], [14, 15], [15, 16]]

        LR = np.array([1, 1, 1, 0, 0, 0, 2, 2, 2, 2, 0, 0, 0, 1, 1, 1], dtype=int) # real left : 0, real right : 1, center : 2

        # color : (blue, green, red)
        lcolor = (255, 0, 0) # blue
        rcolor = (0, 0, 255) # red
        ccolor = (0, 255, 0) # green
        thickness = 3

        for j,c in enumerate(connections):
            start = map(int, kps[c[0]])
            end = map(int, kps[c[1]])
            start = list(start)
            end = list(end)

            if LR[j] == 0:
                color = lcolor
            elif LR[j] == 1:
                color = rcolor
            else:
                color = ccolor

            cv2.line(img, (start[0], start[1]), (end[0], end[1]), color, thickness)
            cv2.circle(img, (start[0], start[1]), thickness=-1, color=(255, 180, 0), radius=3)
            cv2.circle(img, (end[0], end[1]), thickness=-1, color=(255, 180, 0), radius=3)

        return img


    def show2Dpose_gt(self, kps, img):
        connections = [[1, 3], [3, 5], [2, 4], [4, 6], [7, 9], [9, 11], [8, 10], [10, 12]] # 8개

        LR = np.array([2, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1], dtype=int) # real left : 0, real right : 1, center : 2

        # color : (blue, green, red)
        lcolor = (255, 0, 0) # blue
        rcolor = (0, 0, 255) # red
        ccolor = (0, 255, 0) # green
        thickness = 3

        for j,c in enumerate(connections):
            start = map(int, kps[c[0]])
            end = map(int, kps[c[1]])
            start = list(start)
            end = list(end)

            if LR[j] == 0:
                color = lcolor
            elif LR[j] == 1:
                color = rcolor
            else:
                color = ccolor

            cv2.line(img, (start[0], start[1]), (end[0], end[1]), color, thickness)
            cv2.circle(img, (start[0], start[1]), thickness=-1, color=(255, 180, 0), radius=3)
            cv2.circle(img, (end[0], end[1]), thickness=-1, color=(255, 180, 0), radius=3)

        return img

    def visualize(self, images_list, image_shape, model_name, number):
        output_dir1 = f'demo/output/backbone/{model_name}/{self.handed_option}/{number}/{model_name}/'
        output_dir2 = f'demo/output/backbone/{model_name}/{self.handed_option}/{number}/bppc/'
        output_dir3 = f'demo/output/backbone/{model_name}/{self.handed_option}/{number}/sm/'
        output_dir4 = f'demo/output/backbone/{model_name}/{self.handed_option}/{number}/gt/'

        if not os.path.exists(output_dir1):
            os.makedirs(output_dir1)
        if not os.path.exists(output_dir2):
            os.makedirs(output_dir2)
        if not os.path.exists(output_dir3):
            os.makedirs(output_dir3)
        if not os.path.exists(output_dir4):
            os.makedirs(output_dir4)

        gt_keypoint_number = 13
        keypoint_number = 17

        hrnet_kpts = self.hrnet_kpts
        hrnet_kpts = self.hrnet_kpts.clone().detach().cuda()
        bppc_kpts = self.bppc_kpts
        sm_kpts = self.sm_kpts
        if self.gt_kpts is None:
            gt_kpts = self.hrnet_kpts['keypoints'][0]
            gt_kpts = torch.tensor(gt_kpts).cuda()
        if self.gt_kpts is not None:
            gt_kpts = self.gt_kpts
            gt_kpts = torch.tensor(gt_kpts).cuda()

        hrnet_kpts = hrnet_kpts.reshape(-1, keypoint_number, 2)
        bppc_kpts = bppc_kpts.reshape(-1, keypoint_number, 2)
        sm_kpts = sm_kpts.reshape(-1, keypoint_number, 2)
        gt_kpts = gt_kpts.reshape(-1, gt_keypoint_number, 2)

        # print(hrnet_kpts.shape)
        # print(gt_kpts.shape)

        hrnet_kpts = np.array(hrnet_kpts.cpu())*image_shape[:2][::-1]
        bppc_kpts = np.array(bppc_kpts.detach().cpu())*image_shape[:2][::-1]
        sm_kpts = np.array(sm_kpts.detach().cpu())*image_shape[:2][::-1]
        # gt_kpts = np.array(gt_kpts.cpu())*image_shape[:2][::-1]
        # print(hrnet_kpts.shape) # 16, 17, 2, for문 돌려서 show2Dpose 해보기
        # print(bppc_kpts.shape)
        # print(sm_kpts.shape)
        # print(gt_kpts.shape)

        # images = sorted(os.listdir(image_folder), key=self.sort_key)
        # mlb
        # images = sorted(os.listdir(image_folder))[1:]

        for i in range(hrnet_kpts.shape[0]):
            try:
                img = cv2.imread(f'data/images/{self.handed_option}_final/{self.folder_number}/{images_list[i]}')
                # img = cv2.imread(f'data/mlb_images/{number}/{images_list[i]}')
                # print(img)
                # height, width, channels = img.shape

                hrnet_img = self.show2Dpose(hrnet_kpts[i], copy.deepcopy(img))
                bppc_img = self.show2Dpose(bppc_kpts[i], copy.deepcopy(img))
                sm_img = self.show2Dpose(sm_kpts[i], copy.deepcopy(img))
                gt_img = self.show2Dpose_gt(gt_kpts[i], copy.deepcopy(img))

                cv2.imwrite(f'demo/output/backbone/{model_name}/{self.handed_option}/{number}/{model_name}/{i}.png', hrnet_img)
                cv2.imwrite(f'demo/output/backbone/{model_name}/{self.handed_option}/{number}/bppc/{i}.png', bppc_img)
                cv2.imwrite(f'demo/output/backbone/{model_name}/{self.handed_option}/{number}/sm/{i}.png', sm_img)
                cv2.imwrite(f'demo/output/backbone/{model_name}/{self.handed_option}/{number}/gt/{i}.png', gt_img)

                # mlb
                # gt_img = self.show2Dpose(gt_kpts[i], copy.deepcopy(img))
                # cv2.imwrite(f'demo/output/backbone/{model_name}/{number}/gt/{i}.png', gt_img)
                # cv2.imwrite(f'demo/output/hrnet/{i}.png', hrnet_img)
                # cv2.imwrite(f'demo/output/bppc/{i}.png', bppc_img)
                # cv2.imwrite(f'demo/output/sm/{i}.png', sm_img)
                # cv2.imwrite(f'demo/output/backbone/gt/{i}.png', gt_img)
            except IndexError:
                continue