cannot write the results

[cozcinar]

When I run the TestDemo.py, I can not obtain any output.

New video: animal_alpaca01 with 0 frames and size of (0, 0)
Total time for this video 0.000004

[remega]

There may be some mistakes in your opencv. In windows or Mac-OS, a conda installed (pip installed) opencv is fine. But in Linux, you should install it by source code, or it would miss the ffmpeg module, where you can't successfully use the video related apis. Therefore, check your opencv installation, and make sure that you can read/write a video by opencv. Please inform me, if it still doesn't work. 发自我的 iPhone

…

在 2018年1月14日，02:10，CKevin ***@***.***> 写道： When I run the TestDemo.py, I can not obtain any output. New video: animal_alpaca01 with 0 frames and size of (0, 0) Total time for this video 0.000004 — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub, or mute the thread.

[mmejeras]

cozcinar, pls try a different version of Python. I had the same issue with Python 2.7 and running the script on Python 3.6 did the trick.

[AdarshMJ]

Im successfully able to run the code. Thank you for a clean implementation codes. I was wondering whether it is possible to set some threshold to obtain saliency maps? Like for example is it possible to restrict obtaining of saliency for only high motion areas and not all of the video?

[remega]

Actually, in the training process, there is a hyper-parameter \gamma to control the soft restriction of motion. Since the DNN is trained in an end-to-end manner, it could not be hard restricted directly. But, one can conduct a post-processing that mask the motion areas with threshold to the final saliency map. 蒋铼 Lai Jiang PhD Candidate School of Electronic and Information Engineering Beihang University Beijing P.R. China Tel: 86-10-18810446304 Homepage:https://jianglaichina.wixsite.com/laijiang

…

------------------------------------------------------------------ 发件人：adarshmj <notifications@github.com> 发送时间：2018年6月5日(星期二) 11:06 收件人：remega/OMCNN_2CLSTM <OMCNN_2CLSTM@noreply.github.com> 抄　送：Lai Jiang <jianglai.china@aliyun.com>; Comment <comment@noreply.github.com> 主　题：Re: [remega/OMCNN_2CLSTM] cannot write the results (#3) Im successfully able to run the code. Thank you for a clean implementation codes. I was wondering whether it is possible to set some threshold to obtain saliency maps? Like for example is it possible to restrict obtaining of saliency for only high motion areas and not all of the video? — You are receiving this because you commented. Reply to this email directly, view it on GitHub, or mute the thread.