您对华为云开发者网站的整体评价？

非常不满意 非常满意

0

1

2

3

4

5

6

7

8

9

10

*您遇到了哪些问题？（最多选三项）
*您感到满意的原因是？（最多选三项）
*请针对您所遇到的问题给出具体的反馈
200/200
提交 取消
  • 资产集市
  • 教学
  • 实践
  • AI说
  • 案例库
  • 生态合作
  • 专区
我的Gallery
帮助文档
0
汽车零部件的表面缺陷识别
基于目标检测的方法，使用工业相机成像的数据，训练一个汽车零部件表面缺陷识别模型
雨落无痕
40个月以前

    0 0 1528 94
40个月以前
2

2用户已评价

案例内容介绍

随着人们生活水平的提高，我国的汽车消费量是世界上排在前列的国家之一。同样，汽车工业也是支持我国实体经济发展的关键产业，与汽车有关的各种产业也多种多样。汽车行业与我们每个人的人身安全息息有关。汽车零部件的质量是整车质量的基础，汽车零件的缺陷检测可以帮助提高汽车的安全性。然而目前人工目视检测的方式不仅存在效率低下，还可能由于人工疲劳造成缺陷零部件流入行业下流，造成潜在的安全风险。

本案例基于目标检测的方法，使用工业相机成像的数据，训练一个汽车零部件表面缺陷识别模型，可以检测出三种汽车零部件（轴承、火花塞和摇把）表面存在多种缺陷。

注意事项

  1. 本案例推荐使用AI框架：Pytorch-1.4.0；

  2. 本案例需使用 GPU 运行，请查看《ModelArts JupyterLab 硬件规格使用指南》了解切换硬件规格的方法；

  3. 如果您是第一次使用 JupyterLab，请查看《ModelArts JupyterLab使用指导》了解使用方法；

  4. 如果您在使用 JupyterLab 过程中碰到报错，请参考《ModelArts JupyterLab常见问题解决办法》尝试解决问题。

实验步骤

数据集说明

本数据来自工业相机成像。数据集由三种数据构成：轴承、火花塞和摇把。

轴承表面一共存在四种缺陷：

1、连接处缺陷：零件正面两种不同材质连接处出现错误，视为连接处缺陷

2、直角边缺陷：相对于合格零件来说，直角边出现缺口，视为直角边缺陷

3、空洞缺陷（连接处缺陷）：零件正面两种不同材质连接处出现错误，视为连接处缺陷

4、毛刺：零件周围出现不规则凸起，视为毛刺

火花塞一共存在三种缺陷：

1、垂直度问题：瓷下面的接线螺母与火花塞本身的中心线不平行 发生弯曲

2、拔丝异常：螺纹丝纹不完整不顺滑或者遭到破坏

3、间距缺陷：侧电极与中心电极严格的技术要求一般是0.6到1.6毫米

摇把一共存在三种缺陷：

1、颜色缺陷：该零件正常表面为统一色，生产过程中表面出现其他颜色视为颜色缺陷

2、划痕缺陷：相对于合格的汽车塑料零件，摇把表面出现划痕，视为划痕缺陷

3、磨损缺陷：在汽车塑料零件表侧面及外侧部分出现磨损不平整，视为磨损缺陷

1.下载代码和数据集

import moxing as mox
import os

mox.file.copy_parallel('obs://obs-aigallery-zc/notebook/car-parts-detect/','./src')
root_path = './src'
os.chdir(root_path)
INFO:root:Using MoXing-v1.17.3-43fbf97f
INFO:root:Using OBS-Python-SDK-3.20.7

注：本案例采用汽车零部件表面缺陷识别数据集，详细数据集已经放入OBS桶中，路径为obs://obs-aigallery-zc/notebook/car-parts-detect/data。

2.安装依赖库

!cd ./code && pip install -r requirements.txt
Looking in indexes: http://repo.myhuaweicloud.com/repository/pypi/simple
Requirement already satisfied: torch==1.6.0 in /home/ma-user/anaconda3/envs/PyTorch-1.4/lib/python3.7/site-packages (from -r requirements.txt (line 1)) (1.6.0)
Requirement already satisfied: torchvision==0.7.0 in /home/ma-user/anaconda3/envs/PyTorch-1.4/lib/python3.7/site-packages (from -r requirements.txt (line 2)) (0.7.0)
Requirement already satisfied: pycocotools==2.0.2 in /home/ma-user/anaconda3/envs/PyTorch-1.4/lib/python3.7/site-packages (from -r requirements.txt (line 3)) (2.0.2)
Requirement already satisfied: matplotlib>=2.1.0 in /home/ma-user/anaconda3/envs/PyTorch-1.4/lib/python3.7/site-packages (from pycocotools==2.0.2->-r requirements.txt (line 3)) (3.4.3)
Requirement already satisfied: setuptools>=18.0 in /home/ma-user/anaconda3/envs/PyTorch-1.4/lib/python3.7/site-packages (from pycocotools==2.0.2->-r requirements.txt (line 3)) (52.0.0.post20210125)
Requirement already satisfied: cython>=0.27.3 in /home/ma-user/anaconda3/envs/PyTorch-1.4/lib/python3.7/site-packages (from pycocotools==2.0.2->-r requirements.txt (line 3)) (0.27.3)
Requirement already satisfied: numpy in /home/ma-user/anaconda3/envs/PyTorch-1.4/lib/python3.7/site-packages (from torch==1.6.0->-r requirements.txt (line 1)) (1.17.0)
Requirement already satisfied: future in /home/ma-user/anaconda3/envs/PyTorch-1.4/lib/python3.7/site-packages (from torch==1.6.0->-r requirements.txt (line 1)) (0.18.2)
Requirement already satisfied: pillow>=4.1.1 in /home/ma-user/anaconda3/envs/PyTorch-1.4/lib/python3.7/site-packages (from torchvision==0.7.0->-r requirements.txt (line 2)) (6.2.0)
Requirement already satisfied: cycler>=0.10 in /home/ma-user/anaconda3/envs/PyTorch-1.4/lib/python3.7/site-packages (from matplotlib>=2.1.0->pycocotools==2.0.2->-r requirements.txt (line 3)) (0.10.0)
Requirement already satisfied: pyparsing>=2.2.1 in /home/ma-user/anaconda3/envs/PyTorch-1.4/lib/python3.7/site-packages (from matplotlib>=2.1.0->pycocotools==2.0.2->-r requirements.txt (line 3)) (2.4.7)
Requirement already satisfied: python-dateutil>=2.7 in /home/ma-user/anaconda3/envs/PyTorch-1.4/lib/python3.7/site-packages (from matplotlib>=2.1.0->pycocotools==2.0.2->-r requirements.txt (line 3)) (2.8.2)
Requirement already satisfied: kiwisolver>=1.0.1 in /home/ma-user/anaconda3/envs/PyTorch-1.4/lib/python3.7/site-packages (from matplotlib>=2.1.0->pycocotools==2.0.2->-r requirements.txt (line 3)) (1.3.2)
Requirement already satisfied: six in /home/ma-user/anaconda3/envs/PyTorch-1.4/lib/python3.7/site-packages (from cycler>=0.10->matplotlib>=2.1.0->pycocotools==2.0.2->-r requirements.txt (line 3)) (1.16.0)
WARNING: You are using pip version 21.0.1; however, version 21.1.3 is available.
You should consider upgrading via the '/home/ma-user/anaconda3/envs/PyTorch-1.4/bin/python -m pip install --upgrade pip' command.

3.开始训练(本步骤大概需要20分钟)

!python ./code/baseline_released.py \
--data_url ./data \
--train_url ./data/training/ \
--last_path ./submission/model/ \
--train-dir ./ckpt/ \
--ckpt-dir ./code/checkpoints/ \
--num-classes 10 \
--num-epochs 2 \
--batch-size 8
INFO:root:Using MoXing-v1.17.3-43fbf97f
INFO:root:Using OBS-Python-SDK-3.20.7
./data
2021hwsz-data-Annotations.zip  2021hwsz-data-Images.zip  training
./data
./data/training/
./submission/model/
/home/ma-user/anaconda3/envs/PyTorch-1.4/lib/python3.7/site-packages/torchvision/ops/boxes.py:101: UserWarning: This overload of nonzero is deprecated:
	nonzero()
Consider using one of the following signatures instead:
	nonzero(*, bool as_tuple) (Triggered internally at  /pytorch/torch/csrc/utils/python_arg_parser.cpp:766.)
  keep = keep.nonzero().squeeze(1)
Epoch: [0]  [  0/581]  eta: 0:15:37  lr: 0.000027  loss: 4.3435 (4.3435)  loss_classifier: 2.6420 (2.6420)  loss_box_reg: 0.0113 (0.0113)  loss_objectness: 1.6316 (1.6316)  loss_rpn_box_reg: 0.0586 (0.0586)  time: 1.6141  data: 0.5689  max mem: 8738
Epoch: [0]  [ 10/581]  eta: 0:08:25  lr: 0.000199  loss: 3.7142 (3.8453)  loss_classifier: 2.3811 (2.2007)  loss_box_reg: 0.0087 (0.0105)  loss_objectness: 1.1346 (1.5238)  loss_rpn_box_reg: 0.0586 (0.1103)  time: 0.8854  data: 0.0761  max mem: 9004
Epoch: [0]  [ 20/581]  eta: 0:07:55  lr: 0.000372  loss: 1.9897 (2.4596)  loss_classifier: 0.9709 (1.3206)  loss_box_reg: 0.0196 (0.0215)  loss_objectness: 0.6049 (1.0013)  loss_rpn_box_reg: 0.0946 (0.1162)  time: 0.8094  data: 0.0260  max mem: 9214
Epoch: [0]  [ 30/581]  eta: 0:07:40  lr: 0.000544  loss: 0.3589 (1.7705)  loss_classifier: 0.1026 (0.9193)  loss_box_reg: 0.0275 (0.0237)  loss_objectness: 0.1562 (0.7261)  loss_rpn_box_reg: 0.0740 (0.1015)  time: 0.8072  data: 0.0258  max mem: 9214
Epoch: [0]  [ 40/581]  eta: 0:07:27  lr: 0.000716  loss: 0.3433 (1.4644)  loss_classifier: 0.1101 (0.7483)  loss_box_reg: 0.0364 (0.0351)  loss_objectness: 0.1340 (0.5837)  loss_rpn_box_reg: 0.0646 (0.0973)  time: 0.8065  data: 0.0260  max mem: 9214
Epoch: [0]  [ 50/581]  eta: 0:07:16  lr: 0.000888  loss: 0.4946 (1.2727)  loss_classifier: 0.1655 (0.6376)  loss_box_reg: 0.0720 (0.0443)  loss_objectness: 0.1121 (0.4905)  loss_rpn_box_reg: 0.0756 (0.1002)  time: 0.8038  data: 0.0251  max mem: 9214
Epoch: [0]  [ 60/581]  eta: 0:07:06  lr: 0.001061  loss: 0.3955 (1.1218)  loss_classifier: 0.1592 (0.5564)  loss_box_reg: 0.0863 (0.0528)  loss_objectness: 0.0832 (0.4221)  loss_rpn_box_reg: 0.0427 (0.0906)  time: 0.8010  data: 0.0245  max mem: 9214
ignore small bbox < 10 2020-01-14_15_35_18_680.jpg
ignore small bbox < 10 2020-01-14_15_35_28_206.jpg
ignore small bbox < 10 2020-01-14_15_35_59_180.jpg
ignore small bbox < 10 2020-01-14_15_35_59_180.jpg
Epoch: [0]  [ 70/581]  eta: 0:06:57  lr: 0.001233  loss: 0.3567 (1.0225)  loss_classifier: 0.1348 (0.4986)  loss_box_reg: 0.0939 (0.0588)  loss_objectness: 0.0631 (0.3752)  loss_rpn_box_reg: 0.0422 (0.0898)  time: 0.7991  data: 0.0243  max mem: 9214
Epoch: [0]  [ 80/581]  eta: 0:06:48  lr: 0.001405  loss: 0.3457 (0.9366)  loss_classifier: 0.1381 (0.4546)  loss_box_reg: 0.0857 (0.0623)  loss_objectness: 0.0544 (0.3357)  loss_rpn_box_reg: 0.0461 (0.0840)  time: 0.8021  data: 0.0243  max mem: 9214
Epoch: [0]  [ 90/581]  eta: 0:06:39  lr: 0.001577  loss: 0.3125 (0.8667)  loss_classifier: 0.1333 (0.4188)  loss_box_reg: 0.0861 (0.0656)  loss_objectness: 0.0500 (0.3037)  loss_rpn_box_reg: 0.0381 (0.0787)  time: 0.8039  data: 0.0246  max mem: 9214
Epoch: [0]  [100/581]  eta: 0:06:30  lr: 0.001750  loss: 0.2927 (0.8116)  loss_classifier: 0.1201 (0.3893)  loss_box_reg: 0.0820 (0.0673)  loss_objectness: 0.0438 (0.2789)  loss_rpn_box_reg: 0.0378 (0.0761)  time: 0.8018  data: 0.0248  max mem: 9214
Epoch: [0]  [110/581]  eta: 0:06:22  lr: 0.001922  loss: 0.3272 (0.7696)  loss_classifier: 0.1209 (0.3665)  loss_box_reg: 0.1002 (0.0706)  loss_objectness: 0.0442 (0.2578)  loss_rpn_box_reg: 0.0499 (0.0747)  time: 0.8009  data: 0.0247  max mem: 9214
Epoch: [0]  [120/581]  eta: 0:06:13  lr: 0.002094  loss: 0.3272 (0.7363)  loss_classifier: 0.1290 (0.3475)  loss_box_reg: 0.1040 (0.0750)  loss_objectness: 0.0390 (0.2410)  loss_rpn_box_reg: 0.0499 (0.0729)  time: 0.8005  data: 0.0244  max mem: 9214
Epoch: [0]  [130/581]  eta: 0:06:05  lr: 0.002266  loss: 0.3754 (0.7087)  loss_classifier: 0.1245 (0.3309)  loss_box_reg: 0.1102 (0.0780)  loss_objectness: 0.0428 (0.2267)  loss_rpn_box_reg: 0.0625 (0.0731)  time: 0.8001  data: 0.0241  max mem: 9214
Epoch: [0]  [140/581]  eta: 0:05:57  lr: 0.002439  loss: 0.3424 (0.6851)  loss_classifier: 0.1249 (0.3170)  loss_box_reg: 0.1264 (0.0824)  loss_objectness: 0.0526 (0.2139)  loss_rpn_box_reg: 0.0449 (0.0718)  time: 0.8067  data: 0.0247  max mem: 9214
Epoch: [0]  [150/581]  eta: 0:05:53  lr: 0.002611  loss: 0.3424 (0.6685)  loss_classifier: 0.1249 (0.3052)  loss_box_reg: 0.1136 (0.0837)  loss_objectness: 0.0659 (0.2069)  loss_rpn_box_reg: 0.0449 (0.0728)  time: 0.8879  data: 0.0430  max mem: 9268
Epoch: [0]  [160/581]  eta: 0:05:48  lr: 0.002783  loss: 0.3990 (0.6557)  loss_classifier: 0.1165 (0.2939)  loss_box_reg: 0.1083 (0.0866)  loss_objectness: 0.0749 (0.1989)  loss_rpn_box_reg: 0.1080 (0.0764)  time: 0.9598  data: 0.0607  max mem: 9268
Epoch: [0]  [170/581]  eta: 0:05:43  lr: 0.002955  loss: 0.4238 (0.6429)  loss_classifier: 0.1399 (0.2862)  loss_box_reg: 0.1349 (0.0902)  loss_objectness: 0.0574 (0.1900)  loss_rpn_box_reg: 0.0888 (0.0764)  time: 0.9513  data: 0.0606  max mem: 9268
Epoch: [0]  [180/581]  eta: 0:05:36  lr: 0.003128  loss: 0.3991 (0.6284)  loss_classifier: 0.1437 (0.2780)  loss_box_reg: 0.1398 (0.0933)  loss_objectness: 0.0450 (0.1817)  loss_rpn_box_reg: 0.0633 (0.0754)  time: 0.9327  data: 0.0608  max mem: 9268
Epoch: [0]  [190/581]  eta: 0:05:30  lr: 0.003300  loss: 0.3767 (0.6148)  loss_classifier: 0.1175 (0.2691)  loss_box_reg: 0.1628 (0.0975)  loss_objectness: 0.0321 (0.1744)  loss_rpn_box_reg: 0.0364 (0.0738)  time: 0.9208  data: 0.0617  max mem: 9268
Epoch: [0]  [200/581]  eta: 0:05:22  lr: 0.003472  loss: 0.3087 (0.5998)  loss_classifier: 0.1064 (0.2607)  loss_box_reg: 0.1517 (0.0998)  loss_objectness: 0.0247 (0.1673)  loss_rpn_box_reg: 0.0189 (0.0720)  time: 0.9129  data: 0.0623  max mem: 9268
Epoch: [0]  [210/581]  eta: 0:05:15  lr: 0.003644  loss: 0.3108 (0.5904)  loss_classifier: 0.1017 (0.2539)  loss_box_reg: 0.1263 (0.1004)  loss_objectness: 0.0332 (0.1629)  loss_rpn_box_reg: 0.0427 (0.0732)  time: 0.9030  data: 0.0634  max mem: 9268
Epoch: [0]  [220/581]  eta: 0:05:07  lr: 0.003817  loss: 0.3217 (0.5785)  loss_classifier: 0.0964 (0.2465)  loss_box_reg: 0.0995 (0.1000)  loss_objectness: 0.0615 (0.1589)  loss_rpn_box_reg: 0.0556 (0.0732)  time: 0.9082  data: 0.0644  max mem: 9268
Epoch: [0]  [230/581]  eta: 0:04:59  lr: 0.003989  loss: 0.3350 (0.5711)  loss_classifier: 0.1045 (0.2411)  loss_box_reg: 0.0995 (0.1003)  loss_objectness: 0.0690 (0.1561)  loss_rpn_box_reg: 0.0556 (0.0736)  time: 0.9048  data: 0.0647  max mem: 9268
Epoch: [0]  [240/581]  eta: 0:04:51  lr: 0.004161  loss: 0.3350 (0.5594)  loss_classifier: 0.1077 (0.2350)  loss_box_reg: 0.1034 (0.1005)  loss_objectness: 0.0657 (0.1520)  loss_rpn_box_reg: 0.0427 (0.0719)  time: 0.8945  data: 0.0645  max mem: 9268
Epoch: [0]  [250/581]  eta: 0:04:44  lr: 0.004333  loss: 0.2663 (0.5486)  loss_classifier: 0.0975 (0.2302)  loss_box_reg: 0.1060 (0.1012)  loss_objectness: 0.0339 (0.1472)  loss_rpn_box_reg: 0.0223 (0.0701)  time: 0.8987  data: 0.0644  max mem: 9268
Epoch: [0]  [260/581]  eta: 0:04:35  lr: 0.004505  loss: 0.2323 (0.5360)  loss_classifier: 0.0816 (0.2244)  loss_box_reg: 0.1115 (0.1014)  loss_objectness: 0.0228 (0.1422)  loss_rpn_box_reg: 0.0169 (0.0679)  time: 0.9000  data: 0.0641  max mem: 9268
Epoch: [0]  [270/581]  eta: 0:04:27  lr: 0.004678  loss: 0.2115 (0.5263)  loss_classifier: 0.0751 (0.2195)  loss_box_reg: 0.1048 (0.1022)  loss_objectness: 0.0162 (0.1380)  loss_rpn_box_reg: 0.0169 (0.0665)  time: 0.8980  data: 0.0647  max mem: 9268
Epoch: [0]  [280/581]  eta: 0:04:19  lr: 0.004850  loss: 0.2675 (0.5184)  loss_classifier: 0.0939 (0.2155)  loss_box_reg: 0.1329 (0.1034)  loss_objectness: 0.0233 (0.1342)  loss_rpn_box_reg: 0.0214 (0.0653)  time: 0.8937  data: 0.0649  max mem: 9268
Epoch: [0]  [290/581]  eta: 0:04:10  lr: 0.005022  loss: 0.2675 (0.5100)  loss_classifier: 0.0829 (0.2108)  loss_box_reg: 0.1082 (0.1039)  loss_objectness: 0.0233 (0.1304)  loss_rpn_box_reg: 0.0434 (0.0649)  time: 0.8812  data: 0.0642  max mem: 9268
Epoch: [0]  [300/581]  eta: 0:04:02  lr: 0.005194  loss: 0.2382 (0.5000)  loss_classifier: 0.0700 (0.2058)  loss_box_reg: 0.0987 (0.1037)  loss_objectness: 0.0174 (0.1267)  loss_rpn_box_reg: 0.0340 (0.0639)  time: 0.8723  data: 0.0667  max mem: 9268
Epoch: [0]  [310/581]  eta: 0:03:53  lr: 0.005367  loss: 0.2255 (0.4946)  loss_classifier: 0.0682 (0.2018)  loss_box_reg: 0.0994 (0.1039)  loss_objectness: 0.0197 (0.1237)  loss_rpn_box_reg: 0.0318 (0.0651)  time: 0.8676  data: 0.0661  max mem: 9268
Epoch: [0]  [320/581]  eta: 0:03:45  lr: 0.005539  loss: 0.2730 (0.4896)  loss_classifier: 0.0761 (0.1980)  loss_box_reg: 0.1002 (0.1040)  loss_objectness: 0.0246 (0.1207)  loss_rpn_box_reg: 0.0727 (0.0668)  time: 0.8700  data: 0.0649  max mem: 9268
Epoch: [0]  [330/581]  eta: 0:03:36  lr: 0.005711  loss: 0.3171 (0.4846)  loss_classifier: 0.0829 (0.1951)  loss_box_reg: 0.1086 (0.1046)  loss_objectness: 0.0281 (0.1182)  loss_rpn_box_reg: 0.0692 (0.0667)  time: 0.8805  data: 0.0691  max mem: 9268
Epoch: [0]  [340/581]  eta: 0:03:28  lr: 0.005883  loss: 0.2944 (0.4784)  loss_classifier: 0.0949 (0.1921)  loss_box_reg: 0.1086 (0.1048)  loss_objectness: 0.0295 (0.1157)  loss_rpn_box_reg: 0.0344 (0.0657)  time: 0.8840  data: 0.0727  max mem: 9268
Epoch: [0]  [350/581]  eta: 0:03:19  lr: 0.006056  loss: 0.2696 (0.4752)  loss_classifier: 0.1029 (0.1900)  loss_box_reg: 0.1167 (0.1056)  loss_objectness: 0.0275 (0.1143)  loss_rpn_box_reg: 0.0229 (0.0653)  time: 0.8673  data: 0.0673  max mem: 9268
Epoch: [0]  [360/581]  eta: 0:03:10  lr: 0.006228  loss: 0.2696 (0.4700)  loss_classifier: 0.1029 (0.1874)  loss_box_reg: 0.1167 (0.1059)  loss_objectness: 0.0302 (0.1122)  loss_rpn_box_reg: 0.0230 (0.0645)  time: 0.8509  data: 0.0614  max mem: 9268
Epoch: [0]  [370/581]  eta: 0:03:02  lr: 0.006400  loss: 0.2556 (0.4651)  loss_classifier: 0.0789 (0.1844)  loss_box_reg: 0.0995 (0.1058)  loss_objectness: 0.0302 (0.1102)  loss_rpn_box_reg: 0.0479 (0.0647)  time: 0.8469  data: 0.0614  max mem: 9268
Epoch: [0]  [380/581]  eta: 0:02:53  lr: 0.006572  loss: 0.2556 (0.4596)  loss_classifier: 0.0789 (0.1818)  loss_box_reg: 0.0991 (0.1058)  loss_objectness: 0.0279 (0.1080)  loss_rpn_box_reg: 0.0479 (0.0640)  time: 0.8374  data: 0.0601  max mem: 9268
Epoch: [0]  [390/581]  eta: 0:02:44  lr: 0.006745  loss: 0.2351 (0.4545)  loss_classifier: 0.0804 (0.1794)  loss_box_reg: 0.0981 (0.1059)  loss_objectness: 0.0194 (0.1058)  loss_rpn_box_reg: 0.0284 (0.0634)  time: 0.8408  data: 0.0603  max mem: 9268
Epoch: [0]  [400/581]  eta: 0:02:35  lr: 0.006917  loss: 0.2559 (0.4524)  loss_classifier: 0.0854 (0.1779)  loss_box_reg: 0.1202 (0.1071)  loss_objectness: 0.0194 (0.1040)  loss_rpn_box_reg: 0.0385 (0.0634)  time: 0.8503  data: 0.0613  max mem: 9268
Epoch: [0]  [410/581]  eta: 0:02:27  lr: 0.007089  loss: 0.3771 (0.4505)  loss_classifier: 0.1157 (0.1761)  loss_box_reg: 0.1322 (0.1075)  loss_objectness: 0.0337 (0.1027)  loss_rpn_box_reg: 0.0765 (0.0642)  time: 0.8487  data: 0.0609  max mem: 9268
Epoch: [0]  [420/581]  eta: 0:02:18  lr: 0.007261  loss: 0.3816 (0.4484)  loss_classifier: 0.0907 (0.1740)  loss_box_reg: 0.1172 (0.1078)  loss_objectness: 0.0452 (0.1019)  loss_rpn_box_reg: 0.0837 (0.0648)  time: 0.8507  data: 0.0611  max mem: 9268
Epoch: [0]  [430/581]  eta: 0:02:11  lr: 0.007434  loss: 0.3826 (0.4475)  loss_classifier: 0.0930 (0.1741)  loss_box_reg: 0.1118 (0.1078)  loss_objectness: 0.0586 (0.1014)  loss_rpn_box_reg: 0.0416 (0.0641)  time: 1.1058  data: 0.1647  max mem: 9268
Epoch: [0]  [440/581]  eta: 0:02:05  lr: 0.007606  loss: 0.4979 (0.4526)  loss_classifier: 0.2747 (0.1776)  loss_box_reg: 0.1505 (0.1109)  loss_objectness: 0.0579 (0.1007)  loss_rpn_box_reg: 0.0179 (0.0635)  time: 1.4237  data: 0.2378  max mem: 9268
ignore small bbox < 10 2021-05-16-17-01-48.jpg
Epoch: [0]  [450/581]  eta: 0:01:57  lr: 0.007778  loss: 0.6634 (0.4596)  loss_classifier: 0.3065 (0.1810)  loss_box_reg: 0.2350 (0.1147)  loss_objectness: 0.0756 (0.1009)  loss_rpn_box_reg: 0.0211 (0.0629)  time: 1.4947  data: 0.2105  max mem: 9268
Epoch: [0]  [460/581]  eta: 0:01:50  lr: 0.007950  loss: 0.7618 (0.4682)  loss_classifier: 0.3065 (0.1846)  loss_box_reg: 0.3185 (0.1207)  loss_objectness: 0.0819 (0.1006)  loss_rpn_box_reg: 0.0263 (0.0623)  time: 1.5037  data: 0.2171  max mem: 9268
Epoch: [0]  [470/581]  eta: 0:01:42  lr: 0.008123  loss: 0.6012 (0.4700)  loss_classifier: 0.2546 (0.1857)  loss_box_reg: 0.3017 (0.1231)  loss_objectness: 0.0583 (0.0996)  loss_rpn_box_reg: 0.0263 (0.0616)  time: 1.5048  data: 0.2191  max mem: 9268
Epoch: [0]  [480/581]  eta: 0:01:34  lr: 0.008295  loss: 0.5096 (0.4708)  loss_classifier: 0.1917 (0.1856)  loss_box_reg: 0.2443 (0.1261)  loss_objectness: 0.0366 (0.0983)  loss_rpn_box_reg: 0.0219 (0.0608)  time: 1.4995  data: 0.2180  max mem: 9268
Epoch: [0]  [490/581]  eta: 0:01:26  lr: 0.008467  loss: 0.5060 (0.4704)  loss_classifier: 0.1679 (0.1851)  loss_box_reg: 0.2498 (0.1283)  loss_objectness: 0.0305 (0.0969)  loss_rpn_box_reg: 0.0192 (0.0600)  time: 1.5060  data: 0.2204  max mem: 9268
Epoch: [0]  [500/581]  eta: 0:01:17  lr: 0.008639  loss: 0.3778 (0.4688)  loss_classifier: 0.1350 (0.1845)  loss_box_reg: 0.1993 (0.1294)  loss_objectness: 0.0282 (0.0956)  loss_rpn_box_reg: 0.0182 (0.0593)  time: 1.5151  data: 0.2215  max mem: 9268
Epoch: [0]  [510/581]  eta: 0:01:08  lr: 0.008812  loss: 0.4375 (0.4712)  loss_classifier: 0.1651 (0.1865)  loss_box_reg: 0.1535 (0.1299)  loss_objectness: 0.0357 (0.0957)  loss_rpn_box_reg: 0.0172 (0.0591)  time: 1.5165  data: 0.2255  max mem: 9268
Epoch: [0]  [520/581]  eta: 0:00:59  lr: 0.008984  loss: 0.5171 (0.4715)  loss_classifier: 0.2410 (0.1868)  loss_box_reg: 0.1371 (0.1303)  loss_objectness: 0.0660 (0.0952)  loss_rpn_box_reg: 0.0186 (0.0593)  time: 1.5279  data: 0.2338  max mem: 9268
Epoch: [0]  [530/581]  eta: 0:00:50  lr: 0.009156  loss: 0.4152 (0.4705)  loss_classifier: 0.1749 (0.1862)  loss_box_reg: 0.1533 (0.1312)  loss_objectness: 0.0560 (0.0945)  loss_rpn_box_reg: 0.0123 (0.0586)  time: 1.5319  data: 0.2355  max mem: 9268
Epoch: [0]  [540/581]  eta: 0:00:41  lr: 0.009328  loss: 0.3945 (0.4687)  loss_classifier: 0.1424 (0.1851)  loss_box_reg: 0.1940 (0.1323)  loss_objectness: 0.0281 (0.0935)  loss_rpn_box_reg: 0.0110 (0.0579)  time: 1.5247  data: 0.2312  max mem: 9268
Epoch: [0]  [550/581]  eta: 0:00:31  lr: 0.009501  loss: 0.3268 (0.4664)  loss_classifier: 0.0915 (0.1833)  loss_box_reg: 0.1531 (0.1322)  loss_objectness: 0.0423 (0.0928)  loss_rpn_box_reg: 0.0199 (0.0582)  time: 1.5247  data: 0.2297  max mem: 9268
Epoch: [0]  [560/581]  eta: 0:00:21  lr: 0.009673  loss: 0.3885 (0.4672)  loss_classifier: 0.1032 (0.1833)  loss_box_reg: 0.1277 (0.1327)  loss_objectness: 0.0518 (0.0936)  loss_rpn_box_reg: 0.0199 (0.0576)  time: 1.5284  data: 0.2291  max mem: 9268
Epoch: [0]  [570/581]  eta: 0:00:11  lr: 0.009845  loss: 0.4448 (0.4660)  loss_classifier: 0.1628 (0.1828)  loss_box_reg: 0.1553 (0.1333)  loss_objectness: 0.0547 (0.0929)  loss_rpn_box_reg: 0.0097 (0.0570)  time: 1.5305  data: 0.2303  max mem: 9268
Epoch: [0]  [580/581]  eta: 0:00:01  lr: 0.010000  loss: 0.3470 (0.4634)  loss_classifier: 0.1260 (0.1818)  loss_box_reg: 0.1548 (0.1338)  loss_objectness: 0.0200 (0.0916)  loss_rpn_box_reg: 0.0037 (0.0562)  time: 1.4635  data: 0.2236  max mem: 9268
Epoch: [0] Total time: 0:10:02 (1.0368 s / it)
mkdir: ./code/checkpoints/
finish upload ./code/checkpoints/model_0.pth->./data/training/model_0.pth
Epoch: [1]  [  0/581]  eta: 0:13:33  lr: 0.010000  loss: 0.6443 (0.6443)  loss_classifier: 0.3404 (0.3404)  loss_box_reg: 0.0727 (0.0727)  loss_objectness: 0.1873 (0.1873)  loss_rpn_box_reg: 0.0438 (0.0438)  time: 1.4005  data: 0.5803  max mem: 9268
Epoch: [1]  [ 10/581]  eta: 0:08:14  lr: 0.010000  loss: 0.3522 (0.4644)  loss_classifier: 0.1528 (0.1624)  loss_box_reg: 0.0605 (0.0616)  loss_objectness: 0.1210 (0.1747)  loss_rpn_box_reg: 0.0377 (0.0657)  time: 0.8662  data: 0.0787  max mem: 9268
Epoch: [1]  [ 20/581]  eta: 0:07:51  lr: 0.010000  loss: 0.3545 (0.4479)  loss_classifier: 0.1310 (0.1557)  loss_box_reg: 0.0601 (0.0669)  loss_objectness: 0.0860 (0.1389)  loss_rpn_box_reg: 0.0953 (0.0865)  time: 0.8134  data: 0.0281  max mem: 9268
Epoch: [1]  [ 30/581]  eta: 0:07:38  lr: 0.010000  loss: 0.2923 (0.3884)  loss_classifier: 0.0968 (0.1333)  loss_box_reg: 0.0637 (0.0655)  loss_objectness: 0.0643 (0.1122)  loss_rpn_box_reg: 0.0637 (0.0773)  time: 0.8126  data: 0.0279  max mem: 9268
Epoch: [1]  [ 40/581]  eta: 0:07:27  lr: 0.010000  loss: 0.2923 (0.3864)  loss_classifier: 0.1011 (0.1331)  loss_box_reg: 0.0749 (0.0752)  loss_objectness: 0.0582 (0.1007)  loss_rpn_box_reg: 0.0626 (0.0774)  time: 0.8113  data: 0.0281  max mem: 9268
Epoch: [1]  [ 50/581]  eta: 0:07:16  lr: 0.010000  loss: 0.3579 (0.3803)  loss_classifier: 0.1240 (0.1296)  loss_box_reg: 0.1126 (0.0815)  loss_objectness: 0.0517 (0.0898)  loss_rpn_box_reg: 0.0655 (0.0795)  time: 0.8094  data: 0.0273  max mem: 9268
Epoch: [1]  [ 60/581]  eta: 0:07:07  lr: 0.010000  loss: 0.2635 (0.3567)  loss_classifier: 0.0961 (0.1210)  loss_box_reg: 0.0916 (0.0819)  loss_objectness: 0.0373 (0.0805)  loss_rpn_box_reg: 0.0432 (0.0732)  time: 0.8060  data: 0.0264  max mem: 9268
ignore small bbox < 10 2020-01-14_15_35_18_680.jpg
ignore small bbox < 10 2020-01-14_15_35_28_206.jpg
ignore small bbox < 10 2020-01-14_15_35_59_180.jpg
ignore small bbox < 10 2020-01-14_15_35_59_180.jpg
Epoch: [1]  [ 70/581]  eta: 0:06:57  lr: 0.010000  loss: 0.2284 (0.3550)  loss_classifier: 0.0880 (0.1195)  loss_box_reg: 0.0877 (0.0837)  loss_objectness: 0.0358 (0.0765)  loss_rpn_box_reg: 0.0429 (0.0753)  time: 0.8045  data: 0.0261  max mem: 9268
Epoch: [1]  [ 80/581]  eta: 0:06:49  lr: 0.010000  loss: 0.2862 (0.3438)  loss_classifier: 0.1102 (0.1163)  loss_box_reg: 0.0877 (0.0837)  loss_objectness: 0.0338 (0.0713)  loss_rpn_box_reg: 0.0538 (0.0724)  time: 0.8061  data: 0.0260  max mem: 9268
Epoch: [1]  [ 90/581]  eta: 0:06:40  lr: 0.010000  loss: 0.2428 (0.3332)  loss_classifier: 0.0889 (0.1138)  loss_box_reg: 0.0874 (0.0849)  loss_objectness: 0.0283 (0.0662)  loss_rpn_box_reg: 0.0381 (0.0683)  time: 0.8070  data: 0.0264  max mem: 9268
Epoch: [1]  [100/581]  eta: 0:06:31  lr: 0.010000  loss: 0.2306 (0.3250)  loss_classifier: 0.0804 (0.1112)  loss_box_reg: 0.0742 (0.0845)  loss_objectness: 0.0239 (0.0627)  loss_rpn_box_reg: 0.0363 (0.0666)  time: 0.8049  data: 0.0263  max mem: 9268
Epoch: [1]  [110/581]  eta: 0:06:23  lr: 0.010000  loss: 0.2585 (0.3211)  loss_classifier: 0.0866 (0.1103)  loss_box_reg: 0.0931 (0.0863)  loss_objectness: 0.0239 (0.0591)  loss_rpn_box_reg: 0.0460 (0.0655)  time: 0.8044  data: 0.0263  max mem: 9268
Epoch: [1]  [120/581]  eta: 0:06:14  lr: 0.010000  loss: 0.2743 (0.3184)  loss_classifier: 0.0911 (0.1089)  loss_box_reg: 0.1075 (0.0888)  loss_objectness: 0.0194 (0.0565)  loss_rpn_box_reg: 0.0460 (0.0641)  time: 0.8050  data: 0.0269  max mem: 9268
Epoch: [1]  [130/581]  eta: 0:06:06  lr: 0.010000  loss: 0.2798 (0.3161)  loss_classifier: 0.0830 (0.1069)  loss_box_reg: 0.0996 (0.0896)  loss_objectness: 0.0211 (0.0550)  loss_rpn_box_reg: 0.0572 (0.0646)  time: 0.8048  data: 0.0269  max mem: 9268
Epoch: [1]  [140/581]  eta: 0:05:58  lr: 0.010000  loss: 0.2631 (0.3144)  loss_classifier: 0.0817 (0.1057)  loss_box_reg: 0.0964 (0.0916)  loss_objectness: 0.0253 (0.0533)  loss_rpn_box_reg: 0.0479 (0.0638)  time: 0.8101  data: 0.0277  max mem: 9268
Epoch: [1]  [150/581]  eta: 0:05:54  lr: 0.010000  loss: 0.2685 (0.3158)  loss_classifier: 0.0852 (0.1056)  loss_box_reg: 0.0964 (0.0923)  loss_objectness: 0.0332 (0.0530)  loss_rpn_box_reg: 0.0526 (0.0649)  time: 0.8954  data: 0.0478  max mem: 9268
Epoch: [1]  [160/581]  eta: 0:05:50  lr: 0.010000  loss: 0.3625 (0.3195)  loss_classifier: 0.0837 (0.1046)  loss_box_reg: 0.0788 (0.0930)  loss_objectness: 0.0395 (0.0532)  loss_rpn_box_reg: 0.0864 (0.0687)  time: 0.9716  data: 0.0671  max mem: 9268
Epoch: [1]  [170/581]  eta: 0:05:45  lr: 0.010000  loss: 0.3283 (0.3204)  loss_classifier: 0.0998 (0.1045)  loss_box_reg: 0.1127 (0.0949)  loss_objectness: 0.0366 (0.0524)  loss_rpn_box_reg: 0.0851 (0.0687)  time: 0.9635  data: 0.0676  max mem: 9268
Epoch: [1]  [180/581]  eta: 0:05:38  lr: 0.010000  loss: 0.2690 (0.3164)  loss_classifier: 0.0798 (0.1027)  loss_box_reg: 0.1079 (0.0951)  loss_objectness: 0.0326 (0.0510)  loss_rpn_box_reg: 0.0524 (0.0675)  time: 0.9433  data: 0.0675  max mem: 9268
Epoch: [1]  [190/581]  eta: 0:05:32  lr: 0.010000  loss: 0.2349 (0.3114)  loss_classifier: 0.0721 (0.1008)  loss_box_reg: 0.1009 (0.0956)  loss_objectness: 0.0175 (0.0494)  loss_rpn_box_reg: 0.0293 (0.0656)  time: 0.9322  data: 0.0678  max mem: 9268
Epoch: [1]  [200/581]  eta: 0:05:24  lr: 0.010000  loss: 0.2197 (0.3078)  loss_classifier: 0.0636 (0.0993)  loss_box_reg: 0.1009 (0.0961)  loss_objectness: 0.0149 (0.0483)  loss_rpn_box_reg: 0.0177 (0.0641)  time: 0.9235  data: 0.0688  max mem: 9268
Epoch: [1]  [210/581]  eta: 0:05:17  lr: 0.010000  loss: 0.2564 (0.3077)  loss_classifier: 0.0717 (0.0987)  loss_box_reg: 0.1009 (0.0961)  loss_objectness: 0.0261 (0.0477)  loss_rpn_box_reg: 0.0363 (0.0652)  time: 0.9060  data: 0.0683  max mem: 9268
Epoch: [1]  [220/581]  eta: 0:05:09  lr: 0.010000  loss: 0.2774 (0.3074)  loss_classifier: 0.0789 (0.0976)  loss_box_reg: 0.0916 (0.0956)  loss_objectness: 0.0356 (0.0483)  loss_rpn_box_reg: 0.0573 (0.0658)  time: 0.9095  data: 0.0676  max mem: 9268
Epoch: [1]  [230/581]  eta: 0:05:01  lr: 0.010000  loss: 0.2888 (0.3082)  loss_classifier: 0.0857 (0.0974)  loss_box_reg: 0.0973 (0.0962)  loss_objectness: 0.0450 (0.0487)  loss_rpn_box_reg: 0.0505 (0.0659)  time: 0.9066  data: 0.0678  max mem: 9268
Epoch: [1]  [240/581]  eta: 0:04:53  lr: 0.010000  loss: 0.2657 (0.3063)  loss_classifier: 0.0807 (0.0969)  loss_box_reg: 0.1113 (0.0969)  loss_objectness: 0.0394 (0.0482)  loss_rpn_box_reg: 0.0327 (0.0644)  time: 0.8989  data: 0.0687  max mem: 9268
Epoch: [1]  [250/581]  eta: 0:04:45  lr: 0.010000  loss: 0.2414 (0.3035)  loss_classifier: 0.0765 (0.0964)  loss_box_reg: 0.1096 (0.0970)  loss_objectness: 0.0277 (0.0472)  loss_rpn_box_reg: 0.0218 (0.0628)  time: 0.9055  data: 0.0681  max mem: 9268
Epoch: [1]  [260/581]  eta: 0:04:37  lr: 0.010000  loss: 0.1967 (0.2990)  loss_classifier: 0.0681 (0.0951)  loss_box_reg: 0.0954 (0.0972)  loss_objectness: 0.0143 (0.0458)  loss_rpn_box_reg: 0.0138 (0.0609)  time: 0.9042  data: 0.0674  max mem: 9268
Epoch: [1]  [270/581]  eta: 0:04:29  lr: 0.010000  loss: 0.1922 (0.2962)  loss_classifier: 0.0617 (0.0942)  loss_box_reg: 0.0954 (0.0975)  loss_objectness: 0.0111 (0.0449)  loss_rpn_box_reg: 0.0135 (0.0595)  time: 0.9019  data: 0.0679  max mem: 9268
Epoch: [1]  [280/581]  eta: 0:04:20  lr: 0.010000  loss: 0.2217 (0.2947)  loss_classifier: 0.0736 (0.0940)  loss_box_reg: 0.1016 (0.0982)  loss_objectness: 0.0169 (0.0440)  loss_rpn_box_reg: 0.0183 (0.0585)  time: 0.8953  data: 0.0672  max mem: 9268
Epoch: [1]  [290/581]  eta: 0:04:12  lr: 0.010000  loss: 0.2401 (0.2927)  loss_classifier: 0.0694 (0.0931)  loss_box_reg: 0.1053 (0.0984)  loss_objectness: 0.0169 (0.0431)  loss_rpn_box_reg: 0.0366 (0.0581)  time: 0.8824  data: 0.0655  max mem: 9268
Epoch: [1]  [300/581]  eta: 0:04:03  lr: 0.010000  loss: 0.2109 (0.2895)  loss_classifier: 0.0601 (0.0919)  loss_box_reg: 0.0942 (0.0983)  loss_objectness: 0.0095 (0.0420)  loss_rpn_box_reg: 0.0303 (0.0572)  time: 0.8738  data: 0.0652  max mem: 9268
Epoch: [1]  [310/581]  eta: 0:03:54  lr: 0.010000  loss: 0.2141 (0.2902)  loss_classifier: 0.0703 (0.0916)  loss_box_reg: 0.0986 (0.0986)  loss_objectness: 0.0133 (0.0416)  loss_rpn_box_reg: 0.0314 (0.0584)  time: 0.8666  data: 0.0645  max mem: 9268
Epoch: [1]  [320/581]  eta: 0:03:46  lr: 0.010000  loss: 0.2713 (0.2906)  loss_classifier: 0.0708 (0.0909)  loss_box_reg: 0.0986 (0.0986)  loss_objectness: 0.0230 (0.0410)  loss_rpn_box_reg: 0.0660 (0.0600)  time: 0.8691  data: 0.0655  max mem: 9268
Epoch: [1]  [330/581]  eta: 0:03:37  lr: 0.010000  loss: 0.2713 (0.2903)  loss_classifier: 0.0685 (0.0907)  loss_box_reg: 0.0868 (0.0987)  loss_objectness: 0.0235 (0.0407)  loss_rpn_box_reg: 0.0675 (0.0602)  time: 0.8767  data: 0.0698  max mem: 9268
Epoch: [1]  [340/581]  eta: 0:03:29  lr: 0.010000  loss: 0.2550 (0.2891)  loss_classifier: 0.0812 (0.0906)  loss_box_reg: 0.1018 (0.0988)  loss_objectness: 0.0229 (0.0403)  loss_rpn_box_reg: 0.0333 (0.0594)  time: 0.8783  data: 0.0717  max mem: 9268
Epoch: [1]  [350/581]  eta: 0:03:20  lr: 0.010000  loss: 0.2457 (0.2899)  loss_classifier: 0.0896 (0.0910)  loss_box_reg: 0.1069 (0.0995)  loss_objectness: 0.0214 (0.0405)  loss_rpn_box_reg: 0.0240 (0.0590)  time: 0.8628  data: 0.0665  max mem: 9268
Epoch: [1]  [360/581]  eta: 0:03:11  lr: 0.010000  loss: 0.2457 (0.2888)  loss_classifier: 0.0851 (0.0908)  loss_box_reg: 0.1070 (0.0997)  loss_objectness: 0.0231 (0.0400)  loss_rpn_box_reg: 0.0240 (0.0583)  time: 0.8501  data: 0.0618  max mem: 9268
Epoch: [1]  [370/581]  eta: 0:03:02  lr: 0.010000  loss: 0.2535 (0.2884)  loss_classifier: 0.0702 (0.0904)  loss_box_reg: 0.0995 (0.0999)  loss_objectness: 0.0231 (0.0398)  loss_rpn_box_reg: 0.0368 (0.0582)  time: 0.8498  data: 0.0626  max mem: 9268
Epoch: [1]  [380/581]  eta: 0:02:53  lr: 0.010000  loss: 0.2535 (0.2873)  loss_classifier: 0.0705 (0.0902)  loss_box_reg: 0.1048 (0.1002)  loss_objectness: 0.0214 (0.0393)  loss_rpn_box_reg: 0.0439 (0.0576)  time: 0.8408  data: 0.0617  max mem: 9268
Epoch: [1]  [390/581]  eta: 0:02:45  lr: 0.010000  loss: 0.2247 (0.2861)  loss_classifier: 0.0699 (0.0899)  loss_box_reg: 0.1014 (0.1004)  loss_objectness: 0.0157 (0.0387)  loss_rpn_box_reg: 0.0244 (0.0570)  time: 0.8444  data: 0.0623  max mem: 9268
Epoch: [1]  [400/581]  eta: 0:02:36  lr: 0.010000  loss: 0.2173 (0.2870)  loss_classifier: 0.0691 (0.0904)  loss_box_reg: 0.1014 (0.1015)  loss_objectness: 0.0168 (0.0383)  loss_rpn_box_reg: 0.0397 (0.0569)  time: 0.8530  data: 0.0636  max mem: 9268
Epoch: [1]  [410/581]  eta: 0:02:27  lr: 0.010000  loss: 0.3234 (0.2890)  loss_classifier: 0.1057 (0.0909)  loss_box_reg: 0.1389 (0.1023)  loss_objectness: 0.0258 (0.0381)  loss_rpn_box_reg: 0.0687 (0.0576)  time: 0.8518  data: 0.0629  max mem: 9268
Epoch: [1]  [420/581]  eta: 0:02:19  lr: 0.010000  loss: 0.3741 (0.2903)  loss_classifier: 0.0933 (0.0907)  loss_box_reg: 0.1322 (0.1028)  loss_objectness: 0.0340 (0.0386)  loss_rpn_box_reg: 0.0733 (0.0582)  time: 0.8531  data: 0.0636  max mem: 9268
Epoch: [1]  [430/581]  eta: 0:02:12  lr: 0.010000  loss: 0.3426 (0.2916)  loss_classifier: 0.0961 (0.0923)  loss_box_reg: 0.1161 (0.1031)  loss_objectness: 0.0383 (0.0386)  loss_rpn_box_reg: 0.0422 (0.0576)  time: 1.1091  data: 0.1714  max mem: 9268
Epoch: [1]  [440/581]  eta: 0:02:05  lr: 0.010000  loss: 0.4170 (0.2982)  loss_classifier: 0.2368 (0.0967)  loss_box_reg: 0.1587 (0.1061)  loss_objectness: 0.0297 (0.0385)  loss_rpn_box_reg: 0.0149 (0.0569)  time: 1.4198  data: 0.2416  max mem: 9268
ignore small bbox < 10 2021-05-16-17-01-48.jpg
Epoch: [1]  [450/581]  eta: 0:01:58  lr: 0.010000  loss: 0.5809 (0.3038)  loss_classifier: 0.2574 (0.0993)  loss_box_reg: 0.2349 (0.1093)  loss_objectness: 0.0327 (0.0387)  loss_rpn_box_reg: 0.0176 (0.0565)  time: 1.4820  data: 0.2079  max mem: 9268
Epoch: [1]  [460/581]  eta: 0:01:50  lr: 0.010000  loss: 0.5490 (0.3089)  loss_classifier: 0.2095 (0.1014)  loss_box_reg: 0.2681 (0.1128)  loss_objectness: 0.0389 (0.0387)  loss_rpn_box_reg: 0.0237 (0.0559)  time: 1.4890  data: 0.2132  max mem: 9268
Epoch: [1]  [470/581]  eta: 0:01:42  lr: 0.010000  loss: 0.4407 (0.3112)  loss_classifier: 0.1586 (0.1026)  loss_box_reg: 0.2259 (0.1148)  loss_objectness: 0.0310 (0.0386)  loss_rpn_box_reg: 0.0223 (0.0553)  time: 1.4900  data: 0.2186  max mem: 9268
Epoch: [1]  [480/581]  eta: 0:01:34  lr: 0.010000  loss: 0.4024 (0.3131)  loss_classifier: 0.1430 (0.1034)  loss_box_reg: 0.2070 (0.1169)  loss_objectness: 0.0292 (0.0383)  loss_rpn_box_reg: 0.0192 (0.0546)  time: 1.4975  data: 0.2195  max mem: 9268
Epoch: [1]  [490/581]  eta: 0:01:26  lr: 0.010000  loss: 0.4024 (0.3147)  loss_classifier: 0.1268 (0.1040)  loss_box_reg: 0.2243 (0.1189)  loss_objectness: 0.0222 (0.0379)  loss_rpn_box_reg: 0.0178 (0.0538)  time: 1.5010  data: 0.2168  max mem: 9268
Epoch: [1]  [500/581]  eta: 0:01:17  lr: 0.010000  loss: 0.3164 (0.3147)  loss_classifier: 0.1114 (0.1043)  loss_box_reg: 0.1677 (0.1197)  loss_objectness: 0.0179 (0.0376)  loss_rpn_box_reg: 0.0149 (0.0531)  time: 1.5034  data: 0.2182  max mem: 9268
Epoch: [1]  [510/581]  eta: 0:01:09  lr: 0.010000  loss: 0.3600 (0.3181)  loss_classifier: 0.1273 (0.1070)  loss_box_reg: 0.1602 (0.1206)  loss_objectness: 0.0209 (0.0378)  loss_rpn_box_reg: 0.0135 (0.0528)  time: 1.5124  data: 0.2269  max mem: 9268
Epoch: [1]  [520/581]  eta: 0:00:59  lr: 0.010000  loss: 0.4151 (0.3200)  loss_classifier: 0.2156 (0.1084)  loss_box_reg: 0.1488 (0.1211)  loss_objectness: 0.0334 (0.0379)  loss_rpn_box_reg: 0.0176 (0.0527)  time: 1.5179  data: 0.2299  max mem: 9268
Epoch: [1]  [530/581]  eta: 0:00:50  lr: 0.010000  loss: 0.2731 (0.3189)  loss_classifier: 0.0928 (0.1077)  loss_box_reg: 0.1462 (0.1215)  loss_objectness: 0.0273 (0.0377)  loss_rpn_box_reg: 0.0081 (0.0519)  time: 1.5230  data: 0.2286  max mem: 9268
Epoch: [1]  [540/581]  eta: 0:00:41  lr: 0.010000  loss: 0.2554 (0.3182)  loss_classifier: 0.0752 (0.1073)  loss_box_reg: 0.1462 (0.1219)  loss_objectness: 0.0203 (0.0375)  loss_rpn_box_reg: 0.0071 (0.0515)  time: 1.5244  data: 0.2302  max mem: 9268
Epoch: [1]  [550/581]  eta: 0:00:31  lr: 0.010000  loss: 0.2724 (0.3179)  loss_classifier: 0.0712 (0.1065)  loss_box_reg: 0.1069 (0.1214)  loss_objectness: 0.0281 (0.0380)  loss_rpn_box_reg: 0.0174 (0.0519)  time: 1.5213  data: 0.2313  max mem: 9268
Epoch: [1]  [560/581]  eta: 0:00:21  lr: 0.010000  loss: 0.2812 (0.3182)  loss_classifier: 0.0707 (0.1065)  loss_box_reg: 0.0874 (0.1208)  loss_objectness: 0.0668 (0.0391)  loss_rpn_box_reg: 0.0216 (0.0517)  time: 1.5200  data: 0.2313  max mem: 9268
Epoch: [1]  [570/581]  eta: 0:00:11  lr: 0.010000  loss: 0.3402 (0.3182)  loss_classifier: 0.1095 (0.1068)  loss_box_reg: 0.1009 (0.1210)  loss_objectness: 0.0355 (0.0392)  loss_rpn_box_reg: 0.0123 (0.0513)  time: 1.5203  data: 0.2294  max mem: 9268
Epoch: [1]  [580/581]  eta: 0:00:01  lr: 0.010000  loss: 0.2308 (0.3167)  loss_classifier: 0.1049 (0.1067)  loss_box_reg: 0.1009 (0.1207)  loss_objectness: 0.0227 (0.0388)  loss_rpn_box_reg: 0.0042 (0.0505)  time: 1.4580  data: 0.2216  max mem: 9268
Epoch: [1] Total time: 0:10:02 (1.0371 s / it)
finish upload ./code/checkpoints/model_1.pth->./data/training/model_1.pth
upload model_1.pth -> ./submission/model/model_best.pth

4.完成训练，进行测试模型

import json
import inspect
import os
import sys

root_path = os.path.abspath(inspect.getsourcefile(lambda:0))
root_path = '/'.join(root_path.split('/')[:-2])

sys.path.append('./code')

import customize_service as cs

imageService = cs.ImageClassificationService('model_best.pth', './submission/model/model_best.pth')

# 可以把其他预测的图片放在src/image文件夹下，替换下面的文件名称，进行预测
data = imageService.inference("./image/火花塞.jpg")
print(data)
Using GPU for inference
model already
preprocess time: 264.737606048584ms
infer time: 125.19049644470215ms
postprocess time: 0.3173351287841797ms
latency: 390.2454376220703ms
{'result': [{'boxes': [[549.6098022460938, 1353.4212646484375, 856.999267578125, 1626.158203125], [3011.467529296875, 965.0498046875, 3421.206787109375, 1304.65625], [2989.284912109375, 968.4557495117188, 3422.8388671875, 1304.9481201171875], [576.8909301757812, 1340.2587890625, 877.2547607421875, 1652.6400146484375], [2945.8115234375, 964.2322387695312, 3421.817626953125, 1301.81884765625], [1777.4849853515625, 929.34130859375, 2139.784423828125, 1631.8582763671875], [2151.7578125, 1019.7630004882812, 2980.9716796875, 1444.9742431640625], [1126.210693359375, 1270.072998046875, 1245.9949951171875, 1381.199951171875], [2583.181640625, 988.2589721679688, 2990.323974609375, 1420.2467041015625], [3055.580322265625, 990.49560546875, 3369.37646484375, 1279.698974609375]], 'labels': [9, 10, 9, 10, 7, 10, 10, 8, 10, 8], 'scores': [0.8874596953392029, 0.8255313634872437, 0.20651447772979736, 0.181894451379776, 0.1192605122923851, 0.10301008820533752, 0.09060976654291153, 0.05289372429251671, 0.0503094345331192, 0.050185371190309525]}]}

/home/ma-user/anaconda3/envs/PyTorch-1.4/lib/python3.7/site-packages/torchvision/ops/boxes.py:101: UserWarning: This overload of nonzero is deprecated:
	nonzero()
Consider using one of the following signatures instead:
	nonzero(*, bool as_tuple) (Triggered internally at  /pytorch/torch/csrc/utils/python_arg_parser.cpp:766.)
  keep = keep.nonzero().squeeze(1)

至此，本案例结束!

关联资产

汽车零部件表面缺陷识别数据集

2021数字化转型创新应用大赛·创客赛道初赛竞赛数据集

数据集
3268 455 9
aicompetitions

43个月以前

暂无数据

输出样例

文件总大小 44KB
名称
model_best.pth 该文件不支持代码预览
雨落无痕

作者相关案例

汽车零部件的表面缺陷识别
发布于41个月以前

暂无数据

近7天热度

热门案例推荐

使用PPO算法玩超级马里奥兄弟
ModelArts开发者 发布于45个月以前
使用DQN算法玩2048游戏
ModelArts开发者 发布于45个月以前
使用强化学习AlphaZero算法训练五子棋AI
ModelArts开发者 发布于44个月以前
与中国象棋AI对战！
ModelArts开发者 发布于43个月以前
安全帽检测
ModelArts 发布于43个月以前

暂无数据

评论0

2

2人已评价

  • 0%
  • 0%
  • 50%
  • 0%
  • 50%

请登录后评论