PointAD/train.py at master · zqhang/PointAD · 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
169
170
171
172
173
174
175
176
177
178
179
180
import AnomalyCLIP_lib
import torch
import argparse
import torch.nn.functional as F
from prompt_ensemble import AnomalyCLIP_PromptLearner
from loss import FocalLoss, BinaryDiceLoss
from utils import normalize
from dataset import Dataset
from logger import get_logger
from tqdm import tqdm
import numpy as np
import os
import random
from utils import get_transform
def setup_seed(seed):
    torch.manual_seed(seed)
    torch.cuda.manual_seed_all(seed)
    np.random.seed(seed)
    random.seed(seed)
    torch.backends.cudnn.deterministic = True
    torch.backends.cudnn.benchmark = False


def back_to_3d(d2_similarity_map, d2_3d_cor, non_zero_index, ori_resolution = 336):
    # _, h, w, _ = d2_similarity_map.shape
    h = torch.sqrt(torch.tensor(d2_3d_cor.shape[2])).int()
    w = h
    b, nv, num_points, _ = d2_3d_cor.shape
    xx = d2_3d_cor[:, :, :, 0].reshape(-1).long()
    yy = d2_3d_cor[:, :, :, 1].reshape(-1).long()
    nbatch = torch.repeat_interleave(torch.arange(0, b*nv)[:,None], num_points).reshape(-1, ).cuda().long()
    d2_similarity_map = d2_similarity_map.permute(0, 3, 1, 2)
    point_logits = d2_similarity_map[nbatch, :, yy, xx]
    point_logits = point_logits.reshape(b, nv, num_points, 2)
    vweights = torch.ones((1, nv, 1, 1))
    vweights = vweights.reshape(1, -1, 1, 1).to(point_logits.device)
    is_seen = d2_3d_cor[:, :, :, 2].reshape(b, nv, num_points, 1)
    point_logits = point_logits * vweights * is_seen * non_zero_index
    mask = is_seen.bool() | (~non_zero_index.bool())
    point_logits = point_logits.sum(1)/ (mask.sum(1))
    point_logits = point_logits.reshape(b, h, w, 2)
    return point_logits


def train(args):
    logger = get_logger(args.save_path)
    preprocess, target_transform, target_transform_pc = get_transform(args)
    device = "cuda" if torch.cuda.is_available() else "cpu"
    AnomalyCLIP_parameters = {"Prompt_length": args.n_ctx, "learnabel_text_embedding_depth": args.depth, "learnabel_text_embedding_length": args.t_n_ctx}
    model, _ = AnomalyCLIP_lib.load("ViT-L/14@336px", device=device, design_details = AnomalyCLIP_parameters)
    model.eval()
    train_data = Dataset(root=args.train_data_path, transform=preprocess, target_transform=target_transform, target_transform_pc = target_transform_pc, dataset_name = args.dataset, train_dataset_name = args.train_dataset_name, point_size = args.point_size)
    train_dataloader = torch.utils.data.DataLoader(train_data, batch_size=args.batch_size, shuffle=True)
  ##########################################################################################
    prompt_learner = AnomalyCLIP_PromptLearner(model.to("cpu"), AnomalyCLIP_parameters)
    prompt_learner.to(device)
    model.to(device)
    model.visual.DAPM_replace(DPAM_layer = 20)
    ##########################################################################################
    optimizer = torch.optim.Adam(list(prompt_learner.parameters()), lr=args.learning_rate, betas=(0.5, 0.999))

    # losses
    loss_focal = FocalLoss()
    loss_dice = BinaryDiceLoss()

    model.eval()
    prompt_learner.train()
    for epoch in tqdm(range(args.epoch)):
        model.eval()
        prompt_learner.train()
        d2_pixel_loss_list = []
        d3_global_loss_list = []
        d3_point_loss_list = []
        d2_image_loss_list = []
        for items in tqdm(train_dataloader):
            label =  items['anomaly']
            render_image = items['d2_render_img'].to(device)
            b, nv, c, h, w = render_image.shape
            render_image = render_image.reshape(-1, c, h, w)
            d2_render_anomaly = items['d2_render_anomaly'].to(device)
            d2_3d_cor = items['d2_3d_cor'].to(device)
            # mvtec 3d ad has some zero points
            non_zero_index = items['non_zero_index'].to(device)
            non_zero_index = non_zero_index.unsqueeze(1).repeat(1, nv, 1, 1)

            gt = items['img_mask'].squeeze().to(device)
            gt[gt > 0.5] = 1
            gt[gt <= 0.5] = 0


            render_gt_mask = items['d2_render_gt'].to(device)
            render_gt_mask[render_gt_mask > 0.5], render_gt_mask[render_gt_mask <= 0.5] = 1, 0

            render_gt_mask = render_gt_mask.reshape(-1, 1, h, w)

            with torch.no_grad():
                # Apply DPAM to the layer from 6 to 24
                # DPAM_layer represents the number of layer refined by DPAM from top to bottom
                # DPAM_layer = 1, no DPAM is used
                # DPAM_layer = 20 as default
                image_features, patch_features = model.encode_image(render_image, args.features_list, DPAM_layer = 20)
                image_features = image_features / image_features.norm(dim=-1, keepdim=True)

           ####################################
            prompts, tokenized_prompts, compound_prompts_text = prompt_learner(cls_id = None)
            text_features = model.encode_text_learn(prompts, tokenized_prompts, compound_prompts_text).float()
            text_features = torch.stack(torch.chunk(text_features, dim = 0, chunks = 2), dim = 1)
            text_features = text_features/text_features.norm(dim=-1, keepdim=True)
            # Apply DPAM surgery
            text_probs = image_features.unsqueeze(1) @ text_features.permute(0, 2, 1)
            text_probs = text_probs[:, 0, ...]/0.07
            d2_render_anomaly = d2_render_anomaly.reshape(-1,)
            d2_image_loss = F.cross_entropy(text_probs.squeeze(), d2_render_anomaly.long().cuda())
            #########################################################################
            text_probs = torch.chunk(text_probs,  nv, dim = 0)
            text_probs = torch.stack(text_probs, dim = 1).mean(1)
            d3_global_loss = F.cross_entropy(text_probs, label.long().cuda())
            #########################################################################
            # similarity_map_list.append(similarity_map)

            patch_feature = patch_features[0]
            patch_feature = patch_feature/ patch_feature.norm(dim = -1, keepdim = True)
            similarity, _ = AnomalyCLIP_lib.compute_similarity(patch_feature, text_features[0])
            similarity_map = AnomalyCLIP_lib.get_similarity_map(similarity[:, 1:, :], args.image_size).permute(0, 3, 1, 2)
####################################################################
            d3_similarity_map = back_to_3d(similarity_map.permute(0, 2, 3, 1), d2_3d_cor, non_zero_index).permute(0, 3, 1, 2)
####################################################################

            d3_point_loss = 0
            d3_point_loss += loss_dice(d3_similarity_map[:, 1, :, :], gt)
            d3_point_loss += loss_dice(d3_similarity_map[:, 0, :, :], 1-gt)

            d2_pixel_loss = 0
            d2_pixel_loss += loss_focal(similarity_map, render_gt_mask)
            d2_pixel_loss += loss_dice(similarity_map[:, 1, :, :], render_gt_mask)
            d2_pixel_loss += loss_dice(similarity_map[:, 0, :, :], 1-render_gt_mask)


            optimizer.zero_grad()
            (d3_point_loss + d2_pixel_loss + d3_global_loss + d2_image_loss).backward()
            optimizer.step()
            d2_pixel_loss_list.append(d2_pixel_loss.item())
            d3_point_loss_list.append(d3_point_loss.item())
            d3_global_loss_list.append(d3_global_loss.item())
            d2_image_loss_list.append(d2_image_loss.item())
        # logs
        if (epoch + 1) % args.print_freq == 0:
            logger.info('epoch [{}/{}], d2_pixel_loss:{:.4f}, d3_point_loss:{:.4f}, d3_global_loss:{:.4f}, d2_image_loss:{:.4f}'.format(epoch + 1, args.epoch, np.mean(d2_pixel_loss_list), np.mean(d3_point_loss_list), np.mean(d3_global_loss_list), np.mean(d2_image_loss_list)))

        # save model
        if (epoch + 1) % args.save_freq == 0:
            ckp_path = os.path.join(args.save_path, 'epoch_' + str(epoch + 1) + '.pth')
            torch.save({"prompt_learner": prompt_learner.state_dict()}, ckp_path)

if __name__ == '__main__':
    parser = argparse.ArgumentParser("PointAD", add_help=True)
    parser.add_argument("--train_data_path", type=str, default="./data/visa", help="train dataset path")
    parser.add_argument("--save_path", type=str, default='./checkpoint', help='path to save results')


    parser.add_argument("--dataset", type=str, default='mvtec', help="train dataset name")
    parser.add_argument("--train_dataset_name", type=str, default='cookie', help="save frequency")

    parser.add_argument("--depth", type=int, default=9, help="image size")
    parser.add_argument("--n_ctx", type=int, default=12, help="zero shot")
    parser.add_argument("--t_n_ctx", type=int, default=4, help="zero shot")
    parser.add_argument("--feature_map_layer", type=int, nargs="+", default=[0, 1, 2, 3], help="zero shot")
    parser.add_argument("--features_list", type=int, nargs="+", default=[6, 12, 18, 24], help="features used")

    parser.add_argument("--epoch", type=int, default=15, help="epochs")
    parser.add_argument("--learning_rate", type=float, default=0.001, help="learning rate")
    parser.add_argument("--batch_size", type=int, default=8, help="batch size")
    parser.add_argument("--image_size", type=int, default=518, help="image size")
    parser.add_argument("--point_size", type=int, default=336, help="save frequency")
    parser.add_argument("--print_freq", type=int, default=1, help="print frequency")
    parser.add_argument("--save_freq", type=int, default=1, help="save frequency")
    parser.add_argument("--seed", type=int, default=111, help="random seed")
    args = parser.parse_args()
    setup_seed(args.seed)
    train(args)