AST: Asymmetric Student-Teacher Networks for Industrial Anomaly Detection代码运行

最新推荐文章于 2024-10-03 09:02:12 发布
最新推荐文章于 2024-10-03 09:02:12 发布
阅读量349 收藏 2
点赞数 13
分类专栏： 24年实验笔记 文章标签： python 人工智能 深度学习
版权声明：本文为博主原创文章，遵循 CC 4.0 BY-SA 版权协议，转载请附上原文出处链接和本声明。
本文链接：https://blog.csdn.net/tfxzgp/article/details/141263180
版权

环境

設置遠程路徑
在这里插入图片描述
conda create --name zgp_ast python=3.7.7
pip install -r requirements.txt
PIL>=7.1.2改爲Pillow>=7.1.2

Building wheels for collected packages: efficientnet-pytorch
  Building wheel for efficientnet-pytorch (setup.py) ... done
  Created wheel for efficientnet-pytorch: filename=efficientnet_pytorch-0.7.1-py3-none-any.whl size=16429 sha256=6bfca4c176917cd6dd09b95e03683dd0d110a70732cb96c2b4a759aa2b0093ed
  Stored in directory: /home/cszx/.cache/pip/wheels/6e/17/87/3941ed5f9b2bb0cbfa508d1596b68f294d8f1bdec3160ec1a1
Successfully built efficientnet-pytorch
Installing collected packages: urllib3, typing-extensions, tqdm, threadpoolctl, six, pyparsing, Pillow, packaging, nvidia-cuda-runtime-cu11, nvidia-cuda-nvrtc-cu11, nvidia-cublas-cu11, numpy, joblib, idna, fonttools, cycler, charset-normalizer, tifffile, scipy, requests, python-dateutil, nvidia-cudnn-cu11, kiwisolver, torch, scikit-learn, matplotlib, torchvision, efficientnet-pytorch
Successfully installed Pillow-9.5.0 charset-normalizer-3.3.2 cycler-0.11.0 efficientnet-pytorch-0.7.1 fonttools-4.38.0 idna-3.7 joblib-1.3.2 kiwisolver-1.4.5 matplotlib-3.5.3 numpy-1.21.6 nvidia-cublas-cu11-11.10.3.66 nvidia-cuda-nvrtc-cu11-11.7.99 nvidia-cuda-runtime-cu11-11.7.99 nvidia-cudnn-cu11-8.5.0.96 packaging-24.0 pyparsing-3.1.2 python-dateutil-2.9.0.post0 requests-2.31.0 scikit-learn-1.0.2 scipy-1.7.3 six-1.16.0 threadpoolctl-3.1.0 tifffile-2021.11.2 torch-1.13.1 torchvision-0.14.1 tqdm-4.66.5 typing-extensions-4.7.1 urllib3-2.0.7

設置config.py

pre_extracted=True
dataset_dir = ‘/home/cszx/c1/zgp/mvtec_3d_anomaly_detection.tar’

下載
在这里插入图片描述
放到/home/cszx/.cache/torch/hub/checkpoints

preprocess.py

pycharm直接运行
（時間16:53-19:37前）

ssh://cszx@172.29.6.20:22/home/cszx/miniconda3/envs/zgp_ast/bin/python -u /home/cszx/c1/zgp/AST-main/AST-main/preprocess.py
my_experiment
Loaded pretrained weights for efficientnet-b5
cable_gland
100%|███████████████████████████████████████████| 28/28 [00:10<00:00,  2.61it/s]
data/features/cable_gland
100%|█████████████████████████████████████████████| 7/7 [00:04<00:00,  1.46it/s]
data/features/cable_gland
bagel
100%|███████████████████████████████████████████| 31/31 [00:14<00:00,  2.10it/s]
data/features/bagel
100%|█████████████████████████████████████████████| 7/7 [00:06<00:00,  1.08it/s]
data/features/bagel
peach
100%|███████████████████████████████████████████| 46/46 [00:17<00:00,  2.57it/s]
data/features/peach
100%|█████████████████████████████████████████████| 9/9 [00:06<00:00,  1.40it/s]
data/features/peach
carrot
100%|███████████████████████████████████████████| 36/36 [00:16<00:00,  2.16it/s]
data/features/carrot
100%|███████████████████████████████████████████| 10/10 [00:09<00:00,  1.10it/s]
data/features/carrot
dowel
100%|███████████████████████████████████████████| 36/36 [00:12<00:00,  2.88it/s]
data/features/dowel
100%|█████████████████████████████████████████████| 9/9 [00:05<00:00,  1.65it/s]
data/features/dowel
foam
100%|███████████████████████████████████████████| 30/30 [00:15<00:00,  1.96it/s]
data/features/foam
100%|█████████████████████████████████████████████| 7/7 [00:06<00:00,  1.08it/s]
data/features/foam
cookie
100%|███████████████████████████████████████████| 27/27 [00:09<00:00,  2.75it/s]
data/features/cookie
100%|█████████████████████████████████████████████| 9/9 [00:06<00:00,  1.50it/s]
data/features/cookie
rope
100%|███████████████████████████████████████████| 38/38 [00:14<00:00,  2.57it/s]
data/features/rope
100%|█████████████████████████████████████████████| 7/7 [00:04<00:00,  1.42it/s]
data/features/rope
potato
100%|███████████████████████████████████████████| 38/38 [00:17<00:00,  2.19it/s]
data/features/potato
100%|█████████████████████████████████████████████| 8/8 [00:06<00:00,  1.21it/s]
data/features/potato
tire
100%|███████████████████████████████████████████| 27/27 [00:11<00:00,  2.39it/s]
data/features/tire
100%|█████████████████████████████████████████████| 7/7 [00:06<00:00,  1.15it/s]
data/features/tire
cable_gland
	train
		good
100%|█████████████████████████████████████████| 223/223 [04:40<00:00,  1.26s/it]
	test
		cut
100%|███████████████████████████████████████████| 22/22 [00:33<00:00,  1.53s/it]
		good
100%|███████████████████████████████████████████| 21/21 [00:27<00:00,  1.30s/it]
		hole
100%|███████████████████████████████████████████| 22/22 [00:33<00:00,  1.52s/it]
		bent
100%|███████████████████████████████████████████| 21/21 [00:32<00:00,  1.53s/it]
		thread
100%|███████████████████████████████████████████| 22/22 [00:33<00:00,  1.53s/it]
bagel
	train
		good
100%|█████████████████████████████████████████| 244/244 [09:46<00:00,  2.40s/it]
	test
		crack
100%|███████████████████████████████████████████| 22/22 [00:52<00:00,  2.41s/it]
		good
100%|███████████████████████████████████████████| 22/22 [00:51<00:00,  2.34s/it]
		hole
100%|███████████████████████████████████████████| 21/21 [00:49<00:00,  2.35s/it]
		contamination
100%|███████████████████████████████████████████| 22/22 [00:53<00:00,  2.42s/it]
		combined
100%|███████████████████████████████████████████| 23/23 [00:57<00:00,  2.49s/it]
peach
	train
		good
100%|█████████████████████████████████████████| 361/361 [07:31<00:00,  1.25s/it]
	test
		cut
100%|███████████████████████████████████████████| 25/25 [00:30<00:00,  1.23s/it]
		good
100%|███████████████████████████████████████████| 26/26 [00:34<00:00,  1.34s/it]
		hole
100%|███████████████████████████████████████████| 29/29 [00:36<00:00,  1.27s/it]
		contamination
100%|███████████████████████████████████████████| 27/27 [00:33<00:00,  1.22s/it]
		combined
100%|███████████████████████████████████████████| 25/25 [00:32<00:00,  1.28s/it]
carrot
	train
		good
100%|█████████████████████████████████████████| 286/286 [01:09<00:00,  4.09it/s]
	test
		crack
100%|███████████████████████████████████████████| 26/26 [00:06<00:00,  4.23it/s]
		cut
100%|███████████████████████████████████████████| 26/26 [00:06<00:00,  3.81it/s]
		good
100%|███████████████████████████████████████████| 27/27 [00:06<00:00,  3.90it/s]
		hole
100%|███████████████████████████████████████████| 26/26 [00:07<00:00,  3.69it/s]
		contamination
100%|███████████████████████████████████████████| 27/27 [00:06<00:00,  4.35it/s]
		combined
100%|███████████████████████████████████████████| 27/27 [00:07<00:00,  3.76it/s]
dowel
	train
		good
100%|█████████████████████████████████████████| 288/288 [00:13<00:00, 22.15it/s]
	test
		cut
100%|███████████████████████████████████████████| 25/25 [00:01<00:00, 21.07it/s]
		good
100%|███████████████████████████████████████████| 26/26 [00:01<00:00, 22.80it/s]
		bent
100%|███████████████████████████████████████████| 27/27 [00:01<00:00, 21.38it/s]
		contamination
100%|███████████████████████████████████████████| 26/26 [00:01<00:00, 21.81it/s]
		combined
100%|███████████████████████████████████████████| 26/26 [00:01<00:00, 21.38it/s]
foam
	train
		good
100%|█████████████████████████████████████████| 236/236 [04:47<00:00,  1.22s/it]
	test
		cut
100%|███████████████████████████████████████████| 20/20 [00:24<00:00,  1.23s/it]
		good
100%|███████████████████████████████████████████| 20/20 [00:24<00:00,  1.23s/it]
		contamination
100%|███████████████████████████████████████████| 20/20 [00:24<00:00,  1.22s/it]
		combined
100%|███████████████████████████████████████████| 20/20 [00:24<00:00,  1.22s/it]
		color
100%|███████████████████████████████████████████| 20/20 [00:24<00:00,  1.22s/it]
cookie
	train
		good
100%|█████████████████████████████████████████| 210/210 [02:00<00:00,  1.74it/s]
	test
		crack
100%|███████████████████████████████████████████| 27/27 [00:15<00:00,  1.74it/s]
		good
100%|███████████████████████████████████████████| 28/28 [00:14<00:00,  1.97it/s]
		hole
100%|███████████████████████████████████████████| 26/26 [00:15<00:00,  1.65it/s]
		contamination
100%|███████████████████████████████████████████| 25/25 [00:14<00:00,  1.78it/s]
		combined
100%|███████████████████████████████████████████| 25/25 [00:13<00:00,  1.80it/s]
rope
	train
		good
100%|█████████████████████████████████████████| 298/298 [00:18<00:00, 15.82it/s]
	test
		cut
100%|███████████████████████████████████████████| 27/27 [00:01<00:00, 15.76it/s]
		good
100%|███████████████████████████████████████████| 32/32 [00:02<00:00, 14.51it/s]
		open
100%|███████████████████████████████████████████| 17/17 [00:01<00:00, 12.58it/s]
		contamination
100%|███████████████████████████████████████████| 25/25 [00:01<00:00, 15.46it/s]
potato
	train
		good
100%|█████████████████████████████████████████| 300/300 [05:52<00:00,  1.18s/it]
	test
		cut
100%|███████████████████████████████████████████| 23/23 [00:23<00:00,  1.01s/it]
		good
100%|███████████████████████████████████████████| 22/22 [00:30<00:00,  1.41s/it]
		hole
100%|███████████████████████████████████████████| 23/23 [00:23<00:00,  1.02s/it]
		contamination
100%|███████████████████████████████████████████| 23/23 [00:20<00:00,  1.13it/s]
		combined
100%|███████████████████████████████████████████| 23/23 [00:30<00:00,  1.33s/it]
tire
	train
		good
100%|█████████████████████████████████████████| 210/210 [07:42<00:00,  2.20s/it]
	test
		cut
100%|███████████████████████████████████████████| 27/27 [00:59<00:00,  2.20s/it]
		good
100%|███████████████████████████████████████████| 25/25 [00:54<00:00,  2.19s/it]
		hole
100%|███████████████████████████████████████████| 27/27 [00:59<00:00,  2.19s/it]
		contamination
100%|███████████████████████████████████████████| 28/28 [01:01<00:00,  2.21s/it]
		combined
100%|█████████████████████████████████████████████| 5/5 [00:11<00:00,  2.21s/it]

Process finished with exit code 0

train_teacher.py

ssh://cszx@172.29.6.20:22/home/cszx/miniconda3/envs/zgp_ast/bin/python -u /home/cszx/c1/zgp/AST-main/AST-main/train_teacher.py
my_experiment


Train class cable_gland

Train epoch 0
Epoch: 0.0 	 train loss: 1428.7606
Epoch: 0.4 	 train loss: 152.5859
Epoch: 0.8 	 train loss: 142.6143
Epoch: 0.12 	 train loss: 135.0330
Epoch: 0.16 	 train loss: 129.3598
Epoch: 0.20 	 train loss: 125.2070

Compute loss and scores on test set:
Epoch: 0 	 test_loss: 124.6476
AUROC mean over maps: 	 last: 59.17 	 best: 59.17
AUROC  max over maps: 	 last: 68.09 	 best: 68.09

Train epoch 1
Epoch: 1.0 	 train loss: 121.2998
Epoch: 1.4 	 train loss: 117.9424
Epoch: 1.8 	 train loss: 115.0188
Epoch: 1.12 	 train loss: 111.8316
Epoch: 1.16 	 train loss: 108.9810
Epoch: 1.20 	 train loss: 105.7233

Compute loss and scores on test set:
Epoch: 1 	 test_loss: 108.7641
AUROC mean over maps: 	 last: 67.43 	 best: 67.43
AUROC  max over maps: 	 last: 81.83 	 best: 81.83

Train epoch 2
Epoch: 2.0 	 train loss: 102.8643
Epoch: 2.4 	 train loss: 100.1945
Epoch: 2.8 	 train loss: 97.5472
Epoch: 2.12 	 train loss: 95.1442
Epoch: 2.16 	 train loss: 93.1793
Epoch: 2.20 	 train loss: 91.3104

Compute loss and scores on test set:
Epoch: 2 	 test_loss: 98.4839
AUROC mean over maps: 	 last: 76.08 	 best: 76.08
AUROC  max over maps: 	 last: 87.52 	 best: 87.52
student saved


Train class bagel

Train epoch 0
Epoch: 0.0 	 train loss: 1628.9048
Epoch: 0.4 	 train loss: 240.1958
Epoch: 0.8 	 train loss: 207.1606
Epoch: 0.12 	 train loss: 191.1231
Epoch: 0.16 	 train loss: 180.6131
Epoch: 0.20 	 train loss: 172.6140

Compute loss and scores on test set:
Epoch: 0 	 test_loss: 495.8700
AUROC mean over maps: 	 last: 87.24 	 best: 87.24
AUROC  max over maps: 	 last: 98.92 	 best: 98.92

Train epoch 1
Epoch: 1.0 	 train loss: 165.9658
Epoch: 1.4 	 train loss: 159.2075
Epoch: 1.8 	 train loss: 153.7780
Epoch: 1.12 	 train loss: 149.0566
Epoch: 1.16 	 train loss: 144.5442
Epoch: 1.20 	 train loss: 140.2185

Compute loss and scores on test set:
Epoch: 1 	 test_loss: 7545.4372
AUROC mean over maps: 	 last: 91.89 	 best: 91.89
AUROC  max over maps: 	 last: 97.21 	 best: 98.92

Train epoch 2
Epoch: 2.0 	 train loss: 136.3188
Epoch: 2.4 	 train loss: 133.3591
Epoch: 2.8 	 train loss: 129.7109
Epoch: 2.12 	 train loss: 126.7757
Epoch: 2.16 	 train loss: 123.7185
Epoch: 2.20 	 train loss: 121.0764

Compute loss and scores on test set:
Epoch: 2 	 test_loss: 20937.3579
AUROC mean over maps: 	 last: 90.39 	 best: 91.89
AUROC  max over maps: 	 last: 95.25 	 best: 98.92
student saved


Train class peach

Train epoch 0
Epoch: 0.0 	 train loss: 813.2115
Epoch: 0.4 	 train loss: 138.1477
Epoch: 0.8 	 train loss: 119.7025
Epoch: 0.12 	 train loss: 109.7347
Epoch: 0.16 	 train loss: 103.1757
Epoch: 0.20 	 train loss: 97.5845

Compute loss and scores on test set:
Epoch: 0 	 test_loss: 102.9406
AUROC mean over maps: 	 last: 66.84 	 best: 66.84
AUROC  max over maps: 	 last: 96.30 	 best: 96.30

Train epoch 1
Epoch: 1.0 	 train loss: 92.6392
Epoch: 1.4 	 train loss: 87.9831
Epoch: 1.8 	 train loss: 84.3091
Epoch: 1.12 	 train loss: 80.6787
Epoch: 1.16 	 train loss: 77.6049
Epoch: 1.20 	 train loss: 74.7313

Compute loss and scores on test set:
Epoch: 1 	 test_loss: 85.4764
AUROC mean over maps: 	 last: 77.32 	 best: 77.32
AUROC  max over maps: 	 last: 98.51 	 best: 98.51

Train epoch 2
Epoch: 2.0 	 train loss: 72.1046
Epoch: 2.4 	 train loss: 69.7148
Epoch: 2.8 	 train loss: 67.5666
Epoch: 2.12 	 train loss: 65.5892
Epoch: 2.16 	 train loss: 63.6702
Epoch: 2.20 	 train loss: 61.9145

Compute loss and scores on test set:
Epoch: 2 	 test_loss: 76.6544
AUROC mean over maps: 	 last: 82.95 	 best: 82.95
AUROC  max over maps: 	 last: 99.27 	 best: 99.27
student saved


Train class carrot

Train epoch 0
Epoch: 0.0 	 train loss: 510.4417
Epoch: 0.4 	 train loss: 53.7685
Epoch: 0.8 	 train loss: 47.5511
Epoch: 0.12 	 train loss: 43.6778
Epoch: 0.16 	 train loss: 40.3238
Epoch: 0.20 	 train loss: 37.4852

Compute loss and scores on test set:
Epoch: 0 	 test_loss: 44.7910
AUROC mean over maps: 	 last: 71.24 	 best: 71.24
AUROC  max over maps: 	 last: 91.44 	 best: 91.44

Train epoch 1
Epoch: 1.0 	 train loss: 35.1095
Epoch: 1.4 	 train loss: 33.0163
Epoch: 1.8 	 train loss: 31.2473
Epoch: 1.12 	 train loss: 29.7911
Epoch: 1.16 	 train loss: 28.4091
Epoch: 1.20 	 train loss: 26.9724

Compute loss and scores on test set:
Epoch: 1 	 test_loss: 40.3216
AUROC mean over maps: 	 last: 83.33 	 best: 83.33
AUROC  max over maps: 	 last: 96.86 	 best: 96.86

Train epoch 2
Epoch: 2.0 	 train loss: 25.7731
Epoch: 2.4 	 train loss: 24.6105
Epoch: 2.8 	 train loss: 23.5178
Epoch: 2.12 	 train loss: 22.6717
Epoch: 2.16 	 train loss: 21.8394
Epoch: 2.20 	 train loss: 21.1530

Compute loss and scores on test set:
Epoch: 2 	 test_loss: 38.7690
AUROC mean over maps: 	 last: 87.82 	 best: 87.82
AUROC  max over maps: 	 last: 97.73 	 best: 97.73
student saved


Train class dowel

Train epoch 0
Epoch: 0.0 	 train loss: 625.0942
Epoch: 0.4 	 train loss: 65.2493
Epoch: 0.8 	 train loss: 57.8682
Epoch: 0.12 	 train loss: 53.3894
Epoch: 0.16 	 train loss: 49.3961
Epoch: 0.20 	 train loss: 45.6534

Compute loss and scores on test set:
Epoch: 0 	 test_loss: 46.0253
AUROC mean over maps: 	 last: 91.86 	 best: 91.86
AUROC  max over maps: 	 last: 88.83 	 best: 88.83

Train epoch 1
Epoch: 1.0 	 train loss: 42.0591
Epoch: 1.4 	 train loss: 38.4128
Epoch: 1.8 	 train loss: 35.2950
Epoch: 1.12 	 train loss: 32.6539
Epoch: 1.16 	 train loss: 30.4266
Epoch: 1.20 	 train loss: 28.5939

Compute loss and scores on test set:
Epoch: 1 	 test_loss: 35.5890
AUROC mean over maps: 	 last: 95.60 	 best: 95.60
AUROC  max over maps: 	 last: 98.22 	 best: 98.22

Train epoch 2
Epoch: 2.0 	 train loss: 26.9810
Epoch: 2.4 	 train loss: 25.4965
Epoch: 2.8 	 train loss: 24.2059
Epoch: 2.12 	 train loss: 23.0218
Epoch: 2.16 	 train loss: 21.7920
Epoch: 2.20 	 train loss: 20.7449

Compute loss and scores on test set:
Epoch: 2 	 test_loss: 33.1004
AUROC mean over maps: 	 last: 96.38 	 best: 96.38
AUROC  max over maps: 	 last: 98.30 	 best: 98.30
student saved


Train class foam

Train epoch 0
Epoch: 0.0 	 train loss: 1803.6814
Epoch: 0.4 	 train loss: 256.9068
Epoch: 0.8 	 train loss: 237.8092
Epoch: 0.12 	 train loss: 217.6799
Epoch: 0.16 	 train loss: 200.7390
Epoch: 0.20 	 train loss: 189.6523

Compute loss and scores on test set:
Epoch: 0 	 test_loss: 188.3751
AUROC mean over maps: 	 last: 69.62 	 best: 69.62
AUROC  max over maps: 	 last: 81.06 	 best: 81.06

Train epoch 1
Epoch: 1.0 	 train loss: 180.5969
Epoch: 1.4 	 train loss: 172.2224
Epoch: 1.8 	 train loss: 164.5099
Epoch: 1.12 	 train loss: 158.1079
Epoch: 1.16 	 train loss: 152.4902
Epoch: 1.20 	 train loss: 147.9156

Compute loss and scores on test set:
Epoch: 1 	 test_loss: 153.9179
AUROC mean over maps: 	 last: 74.75 	 best: 74.75
AUROC  max over maps: 	 last: 83.38 	 best: 83.38

Train epoch 2
Epoch: 2.0 	 train loss: 143.4146
Epoch: 2.4 	 train loss: 139.5553
Epoch: 2.8 	 train loss: 136.0524
Epoch: 2.12 	 train loss: 132.6063
Epoch: 2.16 	 train loss: 129.6812
Epoch: 2.20 	 train loss: 126.9911

Compute loss and scores on test set:
Epoch: 2 	 test_loss: 136.7932
AUROC mean over maps: 	 last: 76.00 	 best: 76.00
AUROC  max over maps: 	 last: 86.81 	 best: 86.81
student saved


Train class cookie

Train epoch 0
Epoch: 0.0 	 train loss: 1782.8553
Epoch: 0.4 	 train loss: 232.7387
Epoch: 0.8 	 train loss: 200.1909
Epoch: 0.12 	 train loss: 185.4901
Epoch: 0.16 	 train loss: 175.8931
Epoch: 0.20 	 train loss: 167.9045

Compute loss and scores on test set:
Epoch: 0 	 test_loss: 173.7123
AUROC mean over maps: 	 last: 74.93 	 best: 74.93
AUROC  max over maps: 	 last: 96.64 	 best: 96.64

Train epoch 1
Epoch: 1.0 	 train loss: 161.2892
Epoch: 1.4 	 train loss: 155.1882
Epoch: 1.8 	 train loss: 149.2065
Epoch: 1.12 	 train loss: 144.1039
Epoch: 1.16 	 train loss: 139.5222
Epoch: 1.20 	 train loss: 135.0471

Compute loss and scores on test set:
Epoch: 1 	 test_loss: 145.9559
AUROC mean over maps: 	 last: 78.78 	 best: 78.78
AUROC  max over maps: 	 last: 98.20 	 best: 98.20

Train epoch 2
Epoch: 2.0 	 train loss: 130.8761
Epoch: 2.4 	 train loss: 126.9279
Epoch: 2.8 	 train loss: 123.9564
Epoch: 2.12 	 train loss: 120.6194
Epoch: 2.16 	 train loss: 118.3332
Epoch: 2.20 	 train loss: 114.8934

Compute loss and scores on test set:
Epoch: 2 	 test_loss: 130.8849
AUROC mean over maps: 	 last: 83.04 	 best: 83.04
AUROC  max over maps: 	 last: 98.13 	 best: 98.20
student saved


Train class rope

Train epoch 0
Epoch: 0.0 	 train loss: 685.7194
Epoch: 0.4 	 train loss: 78.9130
Epoch: 0.8 	 train loss: 58.3577
Epoch: 0.12 	 train loss: 51.7182
Epoch: 0.16 	 train loss: 46.7987
Epoch: 0.20 	 train loss: 43.4292

Compute loss and scores on test set:
Epoch: 0 	 test_loss: 81.0114
AUROC mean over maps: 	 last: 99.14 	 best: 99.14
AUROC  max over maps: 	 last: 99.73 	 best: 99.73

Train epoch 1
Epoch: 1.0 	 train loss: 40.7731
Epoch: 1.4 	 train loss: 38.4770
Epoch: 1.8 	 train loss: 36.5572
Epoch: 1.12 	 train loss: 34.5402
Epoch: 1.16 	 train loss: 32.5070
Epoch: 1.20 	 train loss: 30.7293

Compute loss and scores on test set:
Epoch: 1 	 test_loss: 104.2694
AUROC mean over maps: 	 last: 99.23 	 best: 99.23
AUROC  max over maps: 	 last: 98.28 	 best: 99.73

Train epoch 2
Epoch: 2.0 	 train loss: 28.8456
Epoch: 2.4 	 train loss: 27.1801
Epoch: 2.8 	 train loss: 25.6590
Epoch: 2.12 	 train loss: 24.2425
Epoch: 2.16 	 train loss: 22.8058
Epoch: 2.20 	 train loss: 21.4199

Compute loss and scores on test set:
Epoch: 2 	 test_loss: 131.7585
AUROC mean over maps: 	 last: 98.96 	 best: 99.23
AUROC  max over maps: 	 last: 97.37 	 best: 99.73
student saved


Train class potato

Train epoch 0
Epoch: 0.0 	 train loss: 575.9104
Epoch: 0.4 	 train loss: 76.3959
Epoch: 0.8 	 train loss: 67.3652
Epoch: 0.12 	 train loss: 61.2599
Epoch: 0.16 	 train loss: 57.5594
Epoch: 0.20 	 train loss: 54.6413

Compute loss and scores on test set:
Epoch: 0 	 test_loss: 59.2487
AUROC mean over maps: 	 last: 53.36 	 best: 53.36
AUROC  max over maps: 	 last: 96.20 	 best: 96.20

Train epoch 1
Epoch: 1.0 	 train loss: 51.9432
Epoch: 1.4 	 train loss: 49.5469
Epoch: 1.8 	 train loss: 47.4526
Epoch: 1.12 	 train loss: 45.3539
Epoch: 1.16 	 train loss: 43.8155
Epoch: 1.20 	 train loss: 41.8991

Compute loss and scores on test set:
Epoch: 1 	 test_loss: 49.1459
AUROC mean over maps: 	 last: 59.49 	 best: 59.49
AUROC  max over maps: 	 last: 97.53 	 best: 97.53

Train epoch 2
Epoch: 2.0 	 train loss: 40.3873
Epoch: 2.4 	 train loss: 39.1562
Epoch: 2.8 	 train loss: 37.8419
Epoch: 2.12 	 train loss: 36.7338
Epoch: 2.16 	 train loss: 35.5961
Epoch: 2.20 	 train loss: 34.6050

Compute loss and scores on test set:
Epoch: 2 	 test_loss: 44.2994
AUROC mean over maps: 	 last: 63.34 	 best: 63.34
AUROC  max over maps: 	 last: 96.64 	 best: 97.53
student saved


Train class tire

Train epoch 0
Epoch: 0.0 	 train loss: 1268.9452
Epoch: 0.4 	 train loss: 165.4561
Epoch: 0.8 	 train loss: 148.5384
Epoch: 0.12 	 train loss: 138.5010
Epoch: 0.16 	 train loss: 130.8241
Epoch: 0.20 	 train loss: 124.2485

Compute loss and scores on test set:
Epoch: 0 	 test_loss: 122.2500
AUROC mean over maps: 	 last: 46.11 	 best: 46.11
AUROC  max over maps: 	 last: 67.22 	 best: 67.22

Train epoch 1
Epoch: 1.0 	 train loss: 118.7098
Epoch: 1.4 	 train loss: 114.0856
Epoch: 1.8 	 train loss: 109.3179
Epoch: 1.12 	 train loss: 105.0755
Epoch: 1.16 	 train loss: 101.0346
Epoch: 1.20 	 train loss: 97.2061

Compute loss and scores on test set:
Epoch: 1 	 test_loss: 100.4169
AUROC mean over maps: 	 last: 56.64 	 best: 56.64
AUROC  max over maps: 	 last: 68.64 	 best: 68.64

Train epoch 2
Epoch: 2.0 	 train loss: 93.9064
Epoch: 2.4 	 train loss: 90.7649
Epoch: 2.8 	 train loss: 87.9450
Epoch: 2.12 	 train loss: 85.4541
Epoch: 2.16 	 train loss: 83.2304
Epoch: 2.20 	 train loss: 81.0927

Compute loss and scores on test set:
Epoch: 2 	 test_loss: 88.3829
AUROC mean over maps: 	 last: 60.05 	 best: 60.05
AUROC  max over maps: 	 last: 68.46 	 best: 68.64
student saved

AUROC % after last epoch
	mean over maps: 81.50 	 max over maps: 92.55
best AUROC %
	mean over maps: 81.68 	 max over maps: 93.26

Process finished with exit code 0

train_student.py

ssh://cszx@172.29.6.20:22/home/cszx/miniconda3/envs/zgp_ast/bin/python -u /home/cszx/c1/zgp/AST-main/AST-main/train_student.py
my_experiment


Train class cable_gland

Train epoch 0
Epoch: 0.0 	 train loss: 0.5756
Epoch: 0.4 	 train loss: 0.2079
Epoch: 0.8 	 train loss: 0.2045
Epoch: 0.12 	 train loss: 0.1780
Epoch: 0.16 	 train loss: 0.1403
Epoch: 0.20 	 train loss: 0.1154

Compute loss and scores on test set:
Epoch: 0 	 test_loss: 0.1613
AUROC mean over maps: 	 last: 85.88 	 best: 85.88
AUROC  max over maps: 	 last: 95.40 	 best: 95.40

Train epoch 1
Epoch: 1.0 	 train loss: 0.0983
Epoch: 1.4 	 train loss: 0.0848
Epoch: 1.8 	 train loss: 0.0742
Epoch: 1.12 	 train loss: 0.0640
Epoch: 1.16 	 train loss: 0.0552
Epoch: 1.20 	 train loss: 0.0478

Compute loss and scores on test set:
Epoch: 1 	 test_loss: 0.1483
AUROC mean over maps: 	 last: 86.81 	 best: 86.81
AUROC  max over maps: 	 last: 95.73 	 best: 95.73

Train epoch 2
Epoch: 2.0 	 train loss: 0.0416
Epoch: 2.4 	 train loss: 0.0362
Epoch: 2.8 	 train loss: 0.0324
Epoch: 2.12 	 train loss: 0.0281
Epoch: 2.16 	 train loss: 0.0254
Epoch: 2.20 	 train loss: 0.0220

Compute loss and scores on test set:
Epoch: 2 	 test_loss: 0.1520
AUROC mean over maps: 	 last: 87.30 	 best: 87.30
AUROC  max over maps: 	 last: 95.57 	 best: 95.73
student saved


Train class bagel

Train epoch 0
Epoch: 0.0 	 train loss: 0.8018
Epoch: 0.4 	 train loss: 0.4100
Epoch: 0.8 	 train loss: 0.4074
Epoch: 0.12 	 train loss: 0.3762
Epoch: 0.16 	 train loss: 0.3177
Epoch: 0.20 	 train loss: 0.2642

Compute loss and scores on test set:
Epoch: 0 	 test_loss: 113.2161
AUROC mean over maps: 	 last: 89.72 	 best: 89.72
AUROC  max over maps: 	 last: 94.27 	 best: 94.27

Train epoch 1
Epoch: 1.0 	 train loss: 0.2218
Epoch: 1.4 	 train loss: 0.1894
Epoch: 1.8 	 train loss: 0.1635
Epoch: 1.12 	 train loss: 0.1425
Epoch: 1.16 	 train loss: 0.1243
Epoch: 1.20 	 train loss: 0.1093

Compute loss and scores on test set:
Epoch: 1 	 test_loss: 113.1789
AUROC mean over maps: 	 last: 91.32 	 best: 91.32
AUROC  max over maps: 	 last: 93.75 	 best: 94.27

Train epoch 2
Epoch: 2.0 	 train loss: 0.0976
Epoch: 2.4 	 train loss: 0.0879
Epoch: 2.8 	 train loss: 0.0789
Epoch: 2.12 	 train loss: 0.0714
Epoch: 2.16 	 train loss: 0.0648
Epoch: 2.20 	 train loss: 0.0588

Compute loss and scores on test set:
Epoch: 2 	 test_loss: 113.1820
AUROC mean over maps: 	 last: 92.15 	 best: 92.15
AUROC  max over maps: 	 last: 93.75 	 best: 94.27
student saved


Train class peach

Train epoch 0
Epoch: 0.0 	 train loss: 0.4312
Epoch: 0.4 	 train loss: 0.2394
Epoch: 0.8 	 train loss: 0.2337
Epoch: 0.12 	 train loss: 0.2090
Epoch: 0.16 	 train loss: 0.1759
Epoch: 0.20 	 train loss: 0.1475

Compute loss and scores on test set:
Epoch: 0 	 test_loss: 0.2279
AUROC mean over maps: 	 last: 87.59 	 best: 87.59
AUROC  max over maps: 	 last: 99.24 	 best: 99.24

Train epoch 1
Epoch: 1.0 	 train loss: 0.1244
Epoch: 1.4 	 train loss: 0.1041
Epoch: 1.8 	 train loss: 0.0861
Epoch: 1.12 	 train loss: 0.0721
Epoch: 1.16 	 train loss: 0.0600
Epoch: 1.20 	 train loss: 0.0509

Compute loss and scores on test set:
Epoch: 1 	 test_loss: 0.2195
AUROC mean over maps: 	 last: 88.13 	 best: 88.13
AUROC  max over maps: 	 last: 99.27 	 best: 99.27

Train epoch 2
Epoch: 2.0 	 train loss: 0.0436
Epoch: 2.4 	 train loss: 0.0374
Epoch: 2.8 	 train loss: 0.0327
Epoch: 2.12 	 train loss: 0.0285
Epoch: 2.16 	 train loss: 0.0253
Epoch: 2.20 	 train loss: 0.0227

Compute loss and scores on test set:
Epoch: 2 	 test_loss: 0.2250
AUROC mean over maps: 	 last: 87.92 	 best: 88.13
AUROC  max over maps: 	 last: 99.24 	 best: 99.27
student saved


Train class carrot

Train epoch 0
Epoch: 0.0 	 train loss: 0.2496
Epoch: 0.4 	 train loss: 0.0871
Epoch: 0.8 	 train loss: 0.0864
Epoch: 0.12 	 train loss: 0.0838
Epoch: 0.16 	 train loss: 0.0737
Epoch: 0.20 	 train loss: 0.0595

Compute loss and scores on test set:
Epoch: 0 	 test_loss: 0.1468
AUROC mean over maps: 	 last: 88.41 	 best: 88.41
AUROC  max over maps: 	 last: 99.19 	 best: 99.19

Train epoch 1
Epoch: 1.0 	 train loss: 0.0462
Epoch: 1.4 	 train loss: 0.0340
Epoch: 1.8 	 train loss: 0.0246
Epoch: 1.12 	 train loss: 0.0180
Epoch: 1.16 	 train loss: 0.0133
Epoch: 1.20 	 train loss: 0.0105

Compute loss and scores on test set:
Epoch: 1 	 test_loss: 0.1464
AUROC mean over maps: 	 last: 88.58 	 best: 88.58
AUROC  max over maps: 	 last: 99.13 	 best: 99.19

Train epoch 2
Epoch: 2.0 	 train loss: 0.0081
Epoch: 2.4 	 train loss: 0.0069
Epoch: 2.8 	 train loss: 0.0059
Epoch: 2.12 	 train loss: 0.0053
Epoch: 2.16 	 train loss: 0.0050
Epoch: 2.20 	 train loss: 0.0043

Compute loss and scores on test set:
Epoch: 2 	 test_loss: 0.1447
AUROC mean over maps: 	 last: 88.61 	 best: 88.61
AUROC  max over maps: 	 last: 99.16 	 best: 99.19
student saved


Train class dowel

Train epoch 0
Epoch: 0.0 	 train loss: 0.2697
Epoch: 0.4 	 train loss: 0.0966
Epoch: 0.8 	 train loss: 0.0960
Epoch: 0.12 	 train loss: 0.0949
Epoch: 0.16 	 train loss: 0.0898
Epoch: 0.20 	 train loss: 0.0741

Compute loss and scores on test set:
Epoch: 0 	 test_loss: 0.1446
AUROC mean over maps: 	 last: 98.56 	 best: 98.56
AUROC  max over maps: 	 last: 98.48 	 best: 98.48

Train epoch 1
Epoch: 1.0 	 train loss: 0.0587
Epoch: 1.4 	 train loss: 0.0451
Epoch: 1.8 	 train loss: 0.0347
Epoch: 1.12 	 train loss: 0.0246
Epoch: 1.16 	 train loss: 0.0182
Epoch: 1.20 	 train loss: 0.0138

Compute loss and scores on test set:
Epoch: 1 	 test_loss: 0.1394
AUROC mean over maps: 	 last: 98.71 	 best: 98.71
AUROC  max over maps: 	 last: 98.37 	 best: 98.48

Train epoch 2
Epoch: 2.0 	 train loss: 0.0108
Epoch: 2.4 	 train loss: 0.0090
Epoch: 2.8 	 train loss: 0.0073
Epoch: 2.12 	 train loss: 0.0067
Epoch: 2.16 	 train loss: 0.0061
Epoch: 2.20 	 train loss: 0.0053

Compute loss and scores on test set:
Epoch: 2 	 test_loss: 0.1378
AUROC mean over maps: 	 last: 98.67 	 best: 98.71
AUROC  max over maps: 	 last: 98.30 	 best: 98.48
student saved


Train class foam

Train epoch 0
Epoch: 0.0 	 train loss: 0.7689
Epoch: 0.4 	 train loss: 0.4029
Epoch: 0.8 	 train loss: 0.4001
Epoch: 0.12 	 train loss: 0.3740
Epoch: 0.16 	 train loss: 0.3271
Epoch: 0.20 	 train loss: 0.2813

Compute loss and scores on test set:
Epoch: 0 	 test_loss: 0.3547
AUROC mean over maps: 	 last: 82.69 	 best: 82.69
AUROC  max over maps: 	 last: 86.06 	 best: 86.06

Train epoch 1
Epoch: 1.0 	 train loss: 0.2412
Epoch: 1.4 	 train loss: 0.2080
Epoch: 1.8 	 train loss: 0.1789
Epoch: 1.12 	 train loss: 0.1568
Epoch: 1.16 	 train loss: 0.1372
Epoch: 1.20 	 train loss: 0.1206

Compute loss and scores on test set:
Epoch: 1 	 test_loss: 0.4004
AUROC mean over maps: 	 last: 83.88 	 best: 83.88
AUROC  max over maps: 	 last: 86.12 	 best: 86.12

Train epoch 2
Epoch: 2.0 	 train loss: 0.1064
Epoch: 2.4 	 train loss: 0.0949
Epoch: 2.8 	 train loss: 0.0839
Epoch: 2.12 	 train loss: 0.0757
Epoch: 2.16 	 train loss: 0.0685
Epoch: 2.20 	 train loss: 0.0604

Compute loss and scores on test set:
Epoch: 2 	 test_loss: 0.3517
AUROC mean over maps: 	 last: 84.00 	 best: 84.00
AUROC  max over maps: 	 last: 85.75 	 best: 86.12
student saved


Train class cookie

Train epoch 0
Epoch: 0.0 	 train loss: 0.7955
Epoch: 0.4 	 train loss: 0.4147
Epoch: 0.8 	 train loss: 0.4100
Epoch: 0.12 	 train loss: 0.3769
Epoch: 0.16 	 train loss: 0.3212
Epoch: 0.20 	 train loss: 0.2724

Compute loss and scores on test set:
Epoch: 0 	 test_loss: 0.3825
AUROC mean over maps: 	 last: 95.15 	 best: 95.15
AUROC  max over maps: 	 last: 100.00 	 best: 100.00

Train epoch 1
Epoch: 1.0 	 train loss: 0.2334
Epoch: 1.4 	 train loss: 0.1996
Epoch: 1.8 	 train loss: 0.1726
,anKHEpoch: 1.12 	 train loss: 0.1489
HHHHHHHHHHHHHHLEpoch: 1.16 	 train loss: 0.1304
Epoch: 1.20 	 train loss: 0.1120

Compute loss and scores on test set:
Epoch: 1 	 test_loss: 0.3212
AUROC mean over maps: 	 last: 95.56 	 best: 95.56
AUROC  max over maps: 	 last: 100.00 	 best: 100.00

Train epoch 2
Epoch: 2.0 	 train loss: 0.0992
Epoch: 2.4 	 train loss: 0.0872
Epoch: 2.8 	 train loss: 0.0777
Epoch: 2.12 	 train loss: 0.0692
Epoch: 2.16 	 train loss: 0.0625
Epoch: 2.20 	 train loss: 0.0554

Compute loss and scores on test set:
Epoch: 2 	 test_loss: 0.3207
AUROC mean over maps: 	 last: 94.66 	 best: 95.56
AUROC  max over maps: 	 last: 99.79 	 best: 100.00
student saved


Train class rope

Train epoch 0
Epoch: 0.0 	 train loss: 0.3141
Epoch: 0.4 	 train loss: 0.1676
Epoch: 0.8 	 train loss: 0.1649
Epoch: 0.12 	 train loss: 0.1554
Epoch: 0.16 	 train loss: 0.1374
Epoch: 0.20 	 train loss: 0.1165

Compute loss and scores on test set:
Epoch: 0 	 test_loss: 0.6708
AUROC mean over maps: 	 last: 98.19 	 best: 98.19
AUROC  max over maps: 	 last: 96.38 	 best: 96.38

Train epoch 1
Epoch: 1.0 	 train loss: 0.0987
Epoch: 1.4 	 train loss: 0.0834
Epoch: 1.8 	 train loss: 0.0700
Epoch: 1.12 	 train loss: 0.0582
Epoch: 1.16 	 train loss: 0.0465
Epoch: 1.20 	 train loss: 0.0374

Compute loss and scores on test set:
Epoch: 1 	 test_loss: 0.6479
AUROC mean over maps: 	 last: 98.46 	 best: 98.46
AUROC  max over maps: 	 last: 96.06 	 best: 96.38

Train epoch 2
Epoch: 2.0 	 train loss: 0.0300
Epoch: 2.4 	 train loss: 0.0233
Epoch: 2.8 	 train loss: 0.0190
Epoch: 2.12 	 train loss: 0.0158
Epoch: 2.16 	 train loss: 0.0132
Epoch: 2.20 	 train loss: 0.0116

Compute loss and scores on test set:
Epoch: 2 	 test_loss: 0.6503
AUROC mean over maps: 	 last: 98.46 	 best: 98.46
AUROC  max over maps: 	 last: 96.06 	 best: 96.38
student saved


Train class potato

Train epoch 0
Epoch: 0.0 	 train loss: 0.2972
Epoch: 0.4 	 train loss: 0.1213
Epoch: 0.8 	 train loss: 0.1207
Epoch: 0.12 	 train loss: 0.1164
Epoch: 0.16 	 train loss: 0.1032
Epoch: 0.20 	 train loss: 0.0856

Compute loss and scores on test set:
Epoch: 0 	 test_loss: 0.1373
AUROC mean over maps: 	 last: 69.96 	 best: 69.96
AUROC  max over maps: 	 last: 99.01 	 best: 99.01

Train epoch 1
Epoch: 1.0 	 train loss: 0.0695
Epoch: 1.4 	 train loss: 0.0553
Epoch: 1.8 	 train loss: 0.0430
Epoch: 1.12 	 train loss: 0.0336
Epoch: 1.16 	 train loss: 0.0264
Epoch: 1.20 	 train loss: 0.0210

Compute loss and scores on test set:
Epoch: 1 	 test_loss: 0.1388
AUROC mean over maps: 	 last: 68.82 	 best: 69.96
AUROC  max over maps: 	 last: 99.16 	 best: 99.16

Train epoch 2
Epoch: 2.0 	 train loss: 0.0173
Epoch: 2.4 	 train loss: 0.0141
Epoch: 2.8 	 train loss: 0.0122
Epoch: 2.12 	 train loss: 0.0108
Epoch: 2.16 	 train loss: 0.0094
Epoch: 2.20 	 train loss: 0.0088

Compute loss and scores on test set:
Epoch: 2 	 test_loss: 0.1393
AUROC mean over maps: 	 last: 69.12 	 best: 69.96
AUROC  max over maps: 	 last: 99.26 	 best: 99.26
student saved


Train class tire

Train epoch 0
Epoch: 0.0 	 train loss: 0.5844
Epoch: 0.4 	 train loss: 0.2577
Epoch: 0.8 	 train loss: 0.2566
Epoch: 0.12 	 train loss: 0.2541
Epoch: 0.16 	 train loss: 0.2445
Epoch: 0.20 	 train loss: 0.2168

Compute loss and scores on test set:
Epoch: 0 	 test_loss: 0.2498
AUROC mean over maps: 	 last: 68.97 	 best: 68.97
AUROC  max over maps: 	 last: 73.93 	 best: 73.93

Train epoch 1
Epoch: 1.0 	 train loss: 0.1854
Epoch: 1.4 	 train loss: 0.1579
Epoch: 1.8 	 train loss: 0.1356
Epoch: 1.12 	 train loss: 0.1166
Epoch: 1.16 	 train loss: 0.0992
Epoch: 1.20 	 train loss: 0.0834

Compute loss and scores on test set:
Epoch: 1 	 test_loss: 0.2078
AUROC mean over maps: 	 last: 71.40 	 best: 71.40
AUROC  max over maps: 	 last: 77.06 	 best: 77.06

Train epoch 2
Epoch: 2.0 	 train loss: 0.0719
Epoch: 2.4 	 train loss: 0.0594
Epoch: 2.8 	 train loss: 0.0514
Epoch: 2.12 	 train loss: 0.0424
Epoch: 2.16 	 train loss: 0.0356
Epoch: 2.20 	 train loss: 0.0309

Compute loss and scores on test set:
Epoch: 2 	 test_loss: 0.2128
AUROC mean over maps: 	 last: 70.57 	 best: 71.40
AUROC  max over maps: 	 last: 76.60 	 best: 77.06
student saved

AUROC % after last epoch
	mean over maps: 87.15 	 max over maps: 94.35
best AUROC %
	mean over maps: 87.43 	 max over maps: 94.58

Process finished with exit code 0

eval.py

ssh://cszx@172.29.6.20:22/home/cszx/miniconda3/envs/zgp_ast/bin/python -u /home/cszx/c1/zgp/AST-main/AST-main/eval.py
my_experiment

Evaluate class cable_gland
AUROC %	mean over maps: 87.30 	 max over maps: 96.44 	 pixel: 97.73

Evaluate class bagel
AUROC %	mean over maps: 92.15 	 max over maps: 93.90 	 pixel: 98.47

Evaluate class peach
AUROC %	mean over maps: 87.92 	 max over maps: 99.27 	 pixel: 99.64

Evaluate class carrot
AUROC %	mean over maps: 88.61 	 max over maps: 99.19 	 pixel: 99.75

Evaluate class dowel
AUROC %	mean over maps: 98.67 	 max over maps: 98.22 	 pixel: 93.05

Evaluate class foam
AUROC %	mean over maps: 84.00 	 max over maps: 86.31 	 pixel: 97.00

Evaluate class cookie
AUROC %	mean over maps: 94.66 	 max over maps: 99.79 	 pixel: 95.53

Evaluate class rope
AUROC %	mean over maps: 98.46 	 max over maps: 96.20 	 pixel: 94.56

Evaluate class potato
AUROC %	mean over maps: 69.12 	 max over maps: 99.31 	 pixel: 99.85

Evaluate class tire
AUROC %	mean over maps: 70.57 	 max over maps: 77.75 	 pixel: 98.79

mean AUROC % over all classes
	mean over maps: 87.15 	 max over maps: 94.64 	 pixel: 97.44

Process finished with exit code 0

后台运行的方法运行报错

nohup: ignoring input
my_experiment
Traceback (most recent call last):
  File "eval.py", line 9, in <module>
    from model import *
  File "/home/cszx/c1/zgp/AST-main/AST-main/model.py", line 5, in <module>
    from efficientnet_pytorch import EfficientNet
ModuleNotFoundError: No module named 'efficientnet_pytorch'

退出所有虚拟环境，再重新激活使用的虚拟环境，取消一下base的影响
在这里插入图片描述
可以了。
在这里插入图片描述

桂萍
关注
专栏目录
Asymmetric Student-Teacher Networks for Industrial Anomaly Detection
毕竟是Shy哥
07-16 861
通过高度不对称的网络作为学生-教师对来避免从教师到学生的非期望的泛化。我们通过引入作为教师的双射规范化流来改进学生-教师网络通过利用师生距离，我们的AST优于教师的密度估计能力。工业缺陷检测，一般使用异常检测的方法，它的可用数据集(带有缺陷的数据集)很少或不完整。这项工作发现了AD的学生-教师方法中以前未知的问题，并提出了一种解决方案，即训练两个神经网络，为无缺陷的训练示例产生相同的输出。师生网络的核心假设是，对于异常情况，两个网络的输出之间的距离更大，因为它们在训练中不存在。
【环境配置】AST: Asymmetric Student-Teacher Networks for Industrial Anomaly Detection
原来也有北瓜
09-23 191
修改源代码中config.py文件的第5行和第6行。（注意这两个目录不要有包含关系，要不然程序会出错）
【阅读笔记】《Student-Teacher Feature Pyramid Matching for Anomaly Detection》
yetzirah的博客
07-13 1355
论文《Student-Teacher Feature Pyramid Matching for Anomaly Detection》的个人阅读笔记
《Student-Teacher Feature Pyramid Matchingfor Anomaly Detection》论文阅读笔记
m0_46314815的博客
01-01 345
作者提出了一种新的特征金字塔匹配技术，并将其纳入到学生-教师异常检测框架中。
Enhancing Diversity in Teacher-Student Networks via Asymmetric branches for Unsupervised reid论文解读
sdsfddfw的博客
01-07 518
收录于WAVC2021 论文地址：https://openaccess.thecvf.com/content/WACV2021/papers/Chen_Enhancing_Diversity_in_Teacher-Student_Networks_via_Asymmetric_Branches_for_Unsupervised_WACV_2021_paper.pdf 出发点 M-T模型可以减小无监督reid任务中聚类的噪声，但是，teacher和student的自我集成容易让网络迅速收敛到一个局部最小值。因
【论文笔记】【CVPR2020】【异常检测 】Uninformed Students: Student–Teacher Anomaly Detection
weixin_44866655的博客
03-03 6163
【论文笔记】【异常检测】Uninformed Students: Student–Teacher Anomaly Detection with Discriminative Latent Embeddings
ANN:Asymmetric Non-local Neural Networks for Semantic Segmentation
qq_36321330的博客
04-11 1402
code pytorch 原始的Non-local Block 结构图 1.Abstract 标准non-local存在的问题： 计算量过大 GPU内存占用过高 作者提出了用于语义分割的非对称非局部神经网络，它具有两个突出的组成部分：非对称金字塔非局部块（APNB：大大减少了计算量和内存消耗）和非对称融合非局部块（AFNB）。 2.Introduction 之前的研究表明： 如果充分利用远...
ANN:Asymmetric Non-local Neural Networks for Semantic Segmentation 文章解读
wildridder的博客
02-04 4335
《Asymmetric Non-local Neural Networks for Semantic Segmentation 》文章解读 文中提出了两个模块Asymmetric Pyramid Non-local Block (APNB) 和Asymmetric Fusion Non-local Block (AFNB)。APNB将金字塔采样利用到non-local块中，在不牺牲性能的情况下大大...
[论文精读]Few-shot domain-adaptive anomaly detection for cross-site brain images
Sherlily的博客
01-16 1067
计算机-人工智能-脑科学与类脑智能
Asymmetric Student-Teacher Networks for Industrial AD
07-16
"Asymmetric Student-Teacher Networks for Industrial AD" 在工业缺陷检测领域，异常检测（AD）是一种常用的方法。但是，传统的学生教师网络（Student-Teacher Networks）方法存在一些问题，如学生和教师网络架构...
Asymmetric Co-Teaching for Unsupervised Cross-Domain
10-19
Asymmetric Co-Teaching for Unsupervised Cross-Domain Person Re-Identification Asymmetric Co-Teaching for Unsupervised Cross-Domain Person Re-Identification 是一种旨在解决跨域人体重识别问题的方法。...
Traffic-knowledge-based relay selection for asymmetric two-way AF relaying networks
02-21
A novel relay selection criterion, in which the traffic-knowledge is utilized, is proposed for asymmetric two-way amplify-and-forward (AF) relaying networks. The system outage probability, diversity ...
注意力机制论文:Asymmetric Non-local Neural Networks for Semantic Segmentation及其PyTorch实现
mingo_敏
06-03 1864
Asymmetric Non-local Neural Networks for Semantic Segmentation PDF: https://arxiv.org/pdf/1908.07678.pdf PyTorch代码: https://github.com/shanglianlm0525/PyTorch-Networks PyTorch代码:
（笔记）第三期书生·浦语大模型实战营（十一卷王场）--书生入门岛通关第2关Python 基础知识
haidizym的博客
10-02 397
学员闯关手册：https://aicarrier.feishu.cn/wiki/ZcgkwqteZi9s4ZkYr0Gcayg1n1g?课程视频：https://www.bilibili.com/video/BV1mS421X7h4/课程文档：https://github.com/InternLM/Tutorial/tree/camp3/docs/L0/Python。
遥感影像-实例分割数据集：iSAID 从切图到YOLO格式数据集制作详细介绍
GIS潮流
09-30 1042
开源数据集isaid标注包含实例分割，但是原始影像太大，很吃显存，一般显卡无法用原始影像直接训练，所以需要对影像进行裁剪，并生成对应的标签，因为想用yolo系列跑模型，所以将标签需要转为txt格式。
酷炫音乐盒: 使用Python和Tkinter配合Pygame打造自己的音乐播放器
宇宙第一小趴菜的博客
09-29 1187
Tkinter是一个功能强大且简单易用的库，适合构建轻量级的GUI应用程序。它在不同的操作系统上具有相同的外观和行为，因此可以实现跨平台的应用程序。当然，Tkinter还有更多的功能和细节，通过阅读官方文档和教程，你可以更深入地了解和学习如何运用Tkinter来创建各种GUI应用程序。
Python | Leetcode Python题解之第448题找到所有数组中消失的数字
Mopes__的博客
10-01 351
Python | Leetcode Python题解之第448题找到所有数组中消失的数字
Python编写的贪吃蛇小游戏
最新发布
WwLK123的博客
10-03 272
Python编写贪吃蛇小游戏。
电子工程术语大全：从1-Wire到ADSL
16- **ADSL (Asymmetric Digital Subscriber Line)**：非对称数字用户线，利用电话线路提供高速下行（通常用于下载）和较慢上行（上传）的数据服务。 这个资源提供的术语列表涵盖了电子工程领域的许多基础和关键...
评论
添加红包

请填写红包祝福语或标题

个

红包个数最小为10个

元

红包金额最低5元

当前余额3.43元 前往充值 >
需支付：10.00元
成就一亿技术人!
领取后你会自动成为博主和红包主的粉丝 规则
hope_wisdom
发出的红包
实付元
使用余额支付
点击重新获取
扫码支付
钱包余额 0

抵扣说明：

1.余额是钱包充值的虚拟货币，按照1:1的比例进行支付金额的抵扣。
2.余额无法直接购买下载，可以购买VIP、付费专栏及课程。

余额充值