luuleitner / deepmtj Goto Github PK

The python installation of deepMTJ on new machines is cumbersome due to cross dependencies of different necessary packages.

Currently, the code uses the ENUM environment with fixed crop sizes for Aixplorer, Telemed, and Esaote. However, for subject studies, these crop sizes may change due to the antoprometry of the participants or the use of gel pads between the lens and the skin.

From a user perspective, what would be most effective to implement at this point? A semi-automatic rectangle to select a crop for each video or video batch? Similar to what we have on our website?

Please let me know your opinions/ideas
Best Christoph

Hi Cristoph! Thanks for sharing your awesome project.

I'm struggling to set up deepMTJ on my computer. In the README, there's a line that says "Create the deepMTJ virtual environment in your Anaconda terminal and install all necessary libraries (listed in the requirements.txt file) [...]". When I cloned the repository, I couldn't find the "requirements.txt" file, so I ran into an error when I ran the code through Anaconda in the deepMTJ directory (i.e. CondaFileIOError: 'requirements.txt'. [Errno 2] No such file or directory: 'requirements.txt'"). Did I miss it somewhere?

I then went to Colab and ran the sample code which worked great. I only had a problem when I went to upload my own video (which is an uncompressed AVI created in MATLAB as I'm not using a clinical-grade ultrasound machine). I tried to match the file parameters to the esaote_video.avi sample video, but it seems I may have missed something. I'm happy to share my video if that would be helpful (though I can't share it here as AVI is unsupported by Github).

Any suggestions you have on either issue would be helpful. Thanks again for sharing!

Lauren