search

StevenLiuWen / ano_pred_cvpr2018

Official implementation of Paper Future Frame Prediction for Anomaly Detection -- A New Baseline, CVPR 2018
433 stars 126 forks source link

the provided shanghaitech model cant reach 0.72 auc #44

Closed tenggyut closed 4 years ago

tenggyut commented 4 years ago

the provided shanghaitech model cant reach 0.72 auc. use the default training setting also cant reach 0.72 auc on shanghaitech dataset.

Any idea?

zachluo commented 4 years ago

image We can reproduce the reported results.

tenggyut commented 4 years ago

Thx your reply! is the result reproduced by the provided model or your own trained model? What's your training and evaluation settings?

zachluo commented 4 years ago

@tenggyut It is reproduced by the provided model. Please refer to the README to get the result.

abhishekaich27 commented 3 years ago

@zachluo how do we set the training for the Shanghai-Tech dataset? the current format is in form of videos. How do you pre-process this?

zzzzyq-18 commented 2 years ago

@tenggyut它是由提供的模型复制的。请参阅 README 以获取结果。

想请问您，shanghaiTech数据集的训练数据，帧率是多少啊