File size: 8,335 Bytes
9afcee2
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
import torch
import torch.nn
import numpy as np
import pdb


class VNL_Loss(torch.nn.Module):
    """
    Virtual Normal Loss Function.
    """
    def __init__(self, focal_x, focal_y, input_size,
                 delta_cos=0.867, delta_diff_x=0.01,
                 delta_diff_y=0.01, delta_diff_z=0.01,
                 delta_z=0.0001, sample_ratio=0.15):
        super(VNL_Loss, self).__init__()
        self.fx = torch.tensor([focal_x], dtype=torch.float32) #.to(cuda0)
        self.fy = torch.tensor([focal_y], dtype=torch.float32) #.to(cuda0)
        self.input_size = input_size
        self.u0 = torch.tensor(input_size[1] // 2, dtype=torch.float32) #.to(cuda0)
        self.v0 = torch.tensor(input_size[0] // 2, dtype=torch.float32) #.to(cuda0)
        self.init_image_coor()
        self.delta_cos = delta_cos
        self.delta_diff_x = delta_diff_x
        self.delta_diff_y = delta_diff_y
        self.delta_diff_z = delta_diff_z
        self.delta_z = delta_z
        self.sample_ratio = sample_ratio

    def init_image_coor(self):
        x_row = np.arange(0, self.input_size[1])
        x = np.tile(x_row, (self.input_size[0], 1))
        x = x[np.newaxis, :, :]
        x = x.astype(np.float32)
        x = torch.from_numpy(x.copy()) #.to(cuda0)
        self.u_u0 = x - self.u0

        y_col = np.arange(0, self.input_size[0])  # y_col = np.arange(0, height)
        y = np.tile(y_col, (self.input_size[1], 1)).T
        y = y[np.newaxis, :, :]
        y = y.astype(np.float32)
        y = torch.from_numpy(y.copy()) #.to(cuda0)
        self.v_v0 = y - self.v0

    def transfer_xyz(self, depth):
        # print('!!!!!!!!!!!!!!!111111 ', self.u_u0.device, torch.abs(depth).device, self.fx.device)
        x = self.u_u0 * torch.abs(depth) / self.fx
        y = self.v_v0 * torch.abs(depth) / self.fy
        z = depth
        pw = torch.cat([x, y, z], 1).permute(0, 2, 3, 1) # [b, h, w, c]
        return pw

    def select_index(self):
        valid_width = self.input_size[1]
        valid_height = self.input_size[0]
        num = valid_width * valid_height
        p1 = np.random.choice(num, int(num * self.sample_ratio), replace=True)
        np.random.shuffle(p1)
        p2 = np.random.choice(num, int(num * self.sample_ratio), replace=True)
        np.random.shuffle(p2)
        p3 = np.random.choice(num, int(num * self.sample_ratio), replace=True)
        np.random.shuffle(p3)

        p1_x = p1 % self.input_size[1]
        p1_y = (p1 / self.input_size[1]).astype(np.int)

        p2_x = p2 % self.input_size[1]
        p2_y = (p2 / self.input_size[1]).astype(np.int)

        p3_x = p3 % self.input_size[1]
        p3_y = (p3 / self.input_size[1]).astype(np.int)
        p123 = {'p1_x': p1_x, 'p1_y': p1_y, 'p2_x': p2_x, 'p2_y': p2_y, 'p3_x': p3_x, 'p3_y': p3_y}
        return p123

    def form_pw_groups(self, p123, pw):
        """
        Form 3D points groups, with 3 points in each grouup.
        :param p123: points index
        :param pw: 3D points
        :return:
        """
        p1_x = p123['p1_x']
        p1_y = p123['p1_y']
        p2_x = p123['p2_x']
        p2_y = p123['p2_y']
        p3_x = p123['p3_x']
        p3_y = p123['p3_y']

        pw1 = pw[:, p1_y, p1_x, :]
        pw2 = pw[:, p2_y, p2_x, :]
        pw3 = pw[:, p3_y, p3_x, :]
        # [B, N, 3(x,y,z), 3(p1,p2,p3)]
        pw_groups = torch.cat([pw1[:, :, :, np.newaxis], pw2[:, :, :, np.newaxis], pw3[:, :, :, np.newaxis]], 3)
        return pw_groups

    def filter_mask(self, p123, gt_xyz, delta_cos=0.867,
                    delta_diff_x=0.005,
                    delta_diff_y=0.005,
                    delta_diff_z=0.005):
        pw = self.form_pw_groups(p123, gt_xyz)
        pw12 = pw[:, :, :, 1] - pw[:, :, :, 0]
        pw13 = pw[:, :, :, 2] - pw[:, :, :, 0]
        pw23 = pw[:, :, :, 2] - pw[:, :, :, 1]
        ###ignore linear
        pw_diff = torch.cat([pw12[:, :, :, np.newaxis], pw13[:, :, :, np.newaxis], pw23[:, :, :, np.newaxis]],
                            3)  # [b, n, 3, 3]
        m_batchsize, groups, coords, index = pw_diff.shape
        proj_query = pw_diff.view(m_batchsize * groups, -1, index).permute(0, 2, 1)  # (B* X CX(3)) [bn, 3(p123), 3(xyz)]
        proj_key = pw_diff.view(m_batchsize * groups, -1, index)  # B X  (3)*C [bn, 3(xyz), 3(p123)]
        q_norm = proj_query.norm(2, dim=2)
        nm = torch.bmm(q_norm.view(m_batchsize * groups, index, 1), q_norm.view(m_batchsize * groups, 1, index)) #[]
        energy = torch.bmm(proj_query, proj_key)  # transpose check [bn, 3(p123), 3(p123)]
        norm_energy = energy / (nm + 1e-8)
        norm_energy = norm_energy.view(m_batchsize * groups, -1)
        mask_cos = torch.sum((norm_energy > delta_cos) + (norm_energy < -delta_cos), 1) > 3  # igonre
        mask_cos = mask_cos.view(m_batchsize, groups)
        ##ignore padding and invilid depth
        mask_pad = torch.sum(pw[:, :, 2, :] > self.delta_z, 2) == 3

        ###ignore near
        mask_x = torch.sum(torch.abs(pw_diff[:, :, 0, :]) < delta_diff_x, 2) > 0
        mask_y = torch.sum(torch.abs(pw_diff[:, :, 1, :]) < delta_diff_y, 2) > 0
        mask_z = torch.sum(torch.abs(pw_diff[:, :, 2, :]) < delta_diff_z, 2) > 0

        mask_ignore = (mask_x & mask_y & mask_z) | mask_cos
        mask_near = ~mask_ignore
        mask = mask_pad & mask_near

        return mask, pw

    def select_points_groups(self, gt_depth, pred_depth):
        pw_gt = self.transfer_xyz(gt_depth)
        pw_pred = self.transfer_xyz(pred_depth)
        #pdb.set_trace()
        B, C, H, W = gt_depth.shape
        p123 = self.select_index()
        # mask:[b, n], pw_groups_gt: [b, n, 3(x,y,z), 3(p1,p2,p3)]
        mask, pw_groups_gt = self.filter_mask(p123, pw_gt,
                                              delta_cos=0.867,
                                              delta_diff_x=0.005,
                                              delta_diff_y=0.005,
                                              delta_diff_z=0.005)

        # [b, n, 3, 3]
        pw_groups_pred = self.form_pw_groups(p123, pw_pred)
        pw_groups_pred[pw_groups_pred[:, :, 2, :] == 0] = 0.0001
        mask_broadcast = mask.repeat(1, 9).reshape(B, 3, 3, -1).permute(0, 3, 1, 2)
        pw_groups_pred_not_ignore = pw_groups_pred[mask_broadcast].reshape(1, -1, 3, 3)
        pw_groups_gt_not_ignore = pw_groups_gt[mask_broadcast].reshape(1, -1, 3, 3)

        return pw_groups_gt_not_ignore, pw_groups_pred_not_ignore

    def forward(self, gt_depth, pred_depth, select=True):
        """
        Virtual normal loss.
        :param pred_depth: predicted depth map, [B,W,H,C]
        :param data: target label, ground truth depth, [B, W, H, C], padding region [padding_up, padding_down]
        :return:
        """
        device = gt_depth.device
        self.fx = self.fx.to(device)
        self.fy = self.fy.to(device)
        self.u0 = self.u0.to(device)
        self.v0 = self.v0.to(device)
        self.u_u0 = self.u_u0.to(device)
        self.v_v0 = self.v_v0.to(device)
        # print("************ ", self.fx.device, self.u_u0.device)

        gt_points, dt_points = self.select_points_groups(gt_depth, pred_depth)

        gt_p12 = gt_points[:, :, :, 1] - gt_points[:, :, :, 0]
        gt_p13 = gt_points[:, :, :, 2] - gt_points[:, :, :, 0]
        dt_p12 = dt_points[:, :, :, 1] - dt_points[:, :, :, 0]
        dt_p13 = dt_points[:, :, :, 2] - dt_points[:, :, :, 0]

        gt_normal = torch.cross(gt_p12, gt_p13, dim=2)
        dt_normal = torch.cross(dt_p12, dt_p13, dim=2)
        dt_norm = torch.norm(dt_normal, 2, dim=2, keepdim=True)
        gt_norm = torch.norm(gt_normal, 2, dim=2, keepdim=True)
        dt_mask = dt_norm == 0.0
        gt_mask = gt_norm == 0.0
        dt_mask = dt_mask.to(torch.float32)
        gt_mask = gt_mask.to(torch.float32)
        dt_mask *= 0.01
        gt_mask *= 0.01
        gt_norm = gt_norm + gt_mask
        dt_norm = dt_norm + dt_mask
        gt_normal = gt_normal / gt_norm
        dt_normal = dt_normal / dt_norm

        #pdb.set_trace()
        loss = torch.abs(gt_normal - dt_normal)
        loss = torch.sum(torch.sum(loss, dim=2), dim=0)
        if select:
            loss, indices = torch.sort(loss, dim=0, descending=False)
            loss = loss[int(loss.size(0) * 0.25):]
        loss = torch.mean(loss)
        return loss