SteelStrip-SurfaceDefect-ImageRecognition
所属分类:模式识别(视觉/语音等)
开发工具:Jupyter Notebook
文件大小:26001KB
下载次数:0
上传日期:2023-03-04 13:28:35
上 传 者:
sh-1993
说明: #美国有线电视新闻网#迁移学习#ResNeSt-50
(#CNN #Transfer Learning #ResNeSt-50)
文件列表:
SteelStrip_Classification_ResNeSt-50.ipynb (110045, 2023-03-04)
test.zip (5365189, 2023-03-04)
train.zip (21380524, 2023-03-04)
# SteelStrip-SurfaceDefect-ImageRecognition
## Dataset Description
Six kinds of typical surface defects of the hot-rolled steel strip are collected.
- Rolled-in scale(氧化皮)
- Patches(斑塊)
- Crazing(龜裂)
- Pitted surface(凹凸不平)
- Inclusion(內含物)
- Scratches(划痕)
![image](https://user-images.githubusercontent.com/77613396/212956373-e6ad537d-5d99-4f47-9214-1c9cb1fc88b3.png)
## 目標是建立影像辨識模型,分類出照片中顯示的缺陷種類
本專案中主要使用 Transfer Learning 借鑑 [zhanghang1***9](https://github.com/zhanghang1***9/ResNeSt) 提出之預訓練模型架構,並在[競賽](https://www.kaggle.com/competitions/nthuieem-hw3-test/leaderboard)的 Public 與 Private 皆達到 100% 準確預測成果。
[AI Defect-Classification by ResNeSt50 程式檔](https://github.com/Kev107034011/SteelStrip-SurfaceDefect-ImageRecognition/blob/main/SteelStrip_Classification_ResNeSt-50.ipynb)
近期下载者:
相关文件:
收藏者: