基于PaddlePaddle复现 Sub-Image Anomaly Detection with Deep Pyramid Correspondences.
SPatially-Adaptive(SPADE) presents an anomaly segmentation approach which does not require a training stage.
It is fast, robust and achieves SOTA on MVTec AD
dataset.
代码由PyTorch翻译至PaddlePaddle,参考@byungjae89_SPADE-pytorch.感谢!
- I used K=5 nearest neighbors, which differs from the original paper K=50.
硬件
- Intel(R) Xeon® CPU E5-2630v4×2
- NVIDIA Tesla V100 32G
- 256 GB RAM
系统环境
-
CentOS 7.3
-
Python 3.6+
-
PaddlePaddle 2.0.1
若你已经下载了 MVTec AD
数据集,请把它移动至 data/mvtec_anomaly_detection.tar.xz
,若没有,训练启动前,脚本将自动下载该数据集。
*Pickle 不能序列化Paddle Tensor,特征学习时无法保存feature map。 (fixed in 2021-04-02)