Multimedia retrieval pipeline for 3d models.
Tested with Python version 3.7. Only tested with Linux due to limited compatibility of the Faiss library with windows.
Create an environment via the environment.yml file by calling the following:
conda env create --name envname --file=environments.yml
Activate environment:
conda activate envname
Install dependencies via requirements.txt by calling the following:
pip install -r requirements.txt
Since Faiss is not installable via pip it has to be compiled from source. This can be done as detailed in the official documentation Both the gpu and cpu versions of Faiss will work.
Make sure appropriate environment is activated. CD into the src folder and call the following:
bokeh serve --show ui_viewer/
This should open as a page in your browser. If not you can use the localhost link that is shown in the terminal.
Using the UI:
- Use the slider to set the number of query results.
- Paste an absolute path to a .ply (non binary) or .off file into the text field located in the top left corner and press enter.
- Wait for stats and graphs to appear.
- Press the green "Visualize results" button to show the models on the canvas.
Which histograms are plotted can be selected via the orange dropdown. Scroll down to see the t-SNE plot. Canvas camera can be rotated by holding left mouse, panned by holding right nad zoomed using the scroll button.
If something isn't working try reloading the page and if that fails restarting the server should do the trick.