特征匹配
https://zhuanlan.zhihu.com/p/52140541
https://www.kaggle.com/c/severstal-steel-defect-detection/discussion/108078#latest-621878

ensemble技巧
https://www.kaggle.com/c/severstal-steel-defect-detection/discussion/107716#latest-624046
https://www.kaggle.com/c/severstal-steel-defect-detection/discussion/111457#latest-642578

這個(gè)鏈接提到訓(xùn)練時(shí)長的問題,或許需要保存中間結(jié)果
https://www.kaggle.com/c/severstal-steel-defect-detection/discussion/108554#latest-626181

提到了Dice-Score
https://www.kaggle.com/c/severstal-steel-defect-detection/discussion/101465#latest-586178

一篇檢測銹斑的論文
https://www.kaggle.com/c/severstal-steel-defect-detection/discussion/101471#latest-625980
https://www.kaggle.com/c/severstal-steel-defect-detection/discussion/109297#latest-631198
https://www.kaggle.com/c/severstal-steel-defect-detection/discussion/108821#latest-629610
https://software.intel.com/en-us/articles/use-machine-learning-to-detect-defects-on-the-steel-surface

引導(dǎo)性鏈接
https://www.kaggle.com/c/severstal-steel-defect-detection/discussion/101969#latest-641353
https://www.kaggle.com/c/severstal-steel-defect-detection/discussion/103296#latest-640460

關(guān)注圖像角落里的第一個(gè)像素的坐標(biāo)到底是(1,1)還是(0,1)
https://www.kaggle.com/c/severstal-steel-defect-detection/discussion/102146#latest-589715

提到了一篇論文討論了語義分割里面的不同類型的loss
https://www.kaggle.com/c/severstal-steel-defect-detection/discussion/102386#latest-625072
https://www.kaggle.com/c/severstal-steel-defect-detection/discussion/110536#latest-639400
https://www.kaggle.com/c/severstal-steel-defect-detection/discussion/108206#latest-635042

提供了一些網(wǎng)絡(luò)
https://www.kaggle.com/c/severstal-steel-defect-detection/discussion/105296#latest-606287

下面這幾個(gè)沒有完全看懂
https://www.kaggle.com/c/severstal-steel-defect-detection/discussion/103861#latest-600125
https://www.kaggle.com/c/severstal-steel-defect-detection/discussion/103367#latest-639821
https://www.kaggle.com/c/severstal-steel-defect-detection/discussion/106477#latest-642453
https://www.kaggle.com/c/severstal-steel-defect-detection/discussion/109423#latest-630712
https://www.kaggle.com/c/severstal-steel-defect-detection/discussion/108270#latest-629664
https://www.kaggle.com/c/severstal-steel-defect-detection/discussion/107889#latest-631449

半監(jiān)督
https://www.kaggle.com/c/severstal-steel-defect-detection/discussion/110426#latest-641084

提到了數(shù)據(jù)增強(qiáng)
https://www.kaggle.com/c/severstal-steel-defect-detection/discussion/104850#latest-606137
https://www.kaggle.com/c/severstal-steel-defect-detection/discussion/109227#latest-640539

貌似是使用了條件隨機(jī)場
https://www.kaggle.com/c/severstal-steel-defect-detection/discussion/106086#latest-613534

蛙哥說先判斷一個(gè)像素是不是銹斑,然后判斷是第幾類
然后提到不要使用所有數(shù)據(jù),那樣反而會(huì)讓得分低下
https://www.kaggle.com/c/severstal-steel-defect-detection/discussion/106099#latest-629814

照片一致,但是標(biāo)簽不一致
https://www.kaggle.com/c/severstal-steel-defect-detection/discussion/107053#latest-621775

pool大小的調(diào)整建議
https://www.kaggle.com/c/severstal-steel-defect-detection/discussion/106952#latest-620343

新手包
https://www.kaggle.com/c/severstal-steel-defect-detection/discussion/106462#latest-641632

說法是34層的resnet最好
https://www.kaggle.com/c/severstal-steel-defect-detection/discussion/108949#latest-636914

以前的語義分割冠軍方案
https://www.kaggle.com/c/siim-acr-pneumothorax-segmentation/discussion/108308#latest-625068

椒鹽噪聲和對(duì)抗驗(yàn)證
https://www.kaggle.com/c/severstal-steel-defect-detection/discussion/111119#latest-640192
https://www.kaggle.com/c/severstal-steel-defect-detection/discussion/106834#latest-633503
https://www.kaggle.com/c/severstal-steel-defect-detection/discussion/108790#latest-627471

找到很多子類
https://www.kaggle.com/c/severstal-steel-defect-detection/discussion/110363#latest-638823

提出一個(gè)問題:
使用預(yù)訓(xùn)練的網(wǎng)絡(luò),但是預(yù)訓(xùn)練的圖片和當(dāng)前的圖片不一樣的時(shí)候如何處理?(帖子內(nèi)容我沒看,其實(shí)就是修改最后一層)
https://www.kaggle.com/c/severstal-steel-defect-detection/discussion/107246#latest-618321

kaggle在語義分割中的得分機(jī)制dice-score
https://www.kaggle.com/c/severstal-steel-defect-detection/discussion/110188#latest-642222

貌似需要扔掉一些圖片
https://www.kaggle.com/c/severstal-steel-defect-detection/discussion/109673#latest-637866

一大堆神經(jīng)網(wǎng)絡(luò)的論文
https://www.kaggle.com/c/severstal-steel-defect-detection/discussion/109370#latest-631305

提到了IOU
https://www.kaggle.com/c/severstal-steel-defect-detection/discussion/109847#latest-632505

語義分割網(wǎng)絡(luò)回顧
https://www.kaggle.com/c/severstal-steel-defect-detection/discussion/109318#latest-629292

下面這個(gè)似乎非常重要，據(jù)說只要移除False Positive，就可以獲得0.9117
https://www.kaggle.com/evgenyshtepin/severstal-mlcomp-catalyst-infer-0-90726
https://www.kaggle.com/c/severstal-steel-defect-detection/discussion/106462#latest-634450

這個(gè)EDA做的很漂亮
https://www.kaggle.com/avirald/clear-mask-visualization-and-simple-eda

這個(gè)鏈接提到IoU是一種 loss
https://www.kaggle.com/rishabhiitbhu/unet-starter-kernel-pytorch-lb-0-88

總結(jié)

以上是生活随笔為你收集整理的Severstal: Steel Defect Detection比赛的discussion调研的全部內(nèi)容，希望文章能夠幫你解決所遇到的問題。

如果覺得生活随笔網(wǎng)站內(nèi)容還不錯(cuò)，歡迎將生活随笔推薦給好友。