How can i do inference from NYUD trained model using python?

[athiselvam]

Hi @yun-liu ,

I've downloaded NYUD trained model. I need to get edge image for my own house image using python. How can I get this done?

Thanks for your attention.

[yun-liu]

If your image has RGBD data, you can refer to the script of RCF-singlescale-NYUD.ipynb to compute edges. Note that the depth data is encoded into HHA using the algorithm in the paper of Learning Rich Features from RGB-D Images for Object Detection and Segmentation. If your image only has RGB data, you can refer to the script of RCF-singlescale.ipynb to compute edges.

[athiselvam]

I wanted to infer from NUYD trained model. Because my test images are house images. But test images don't have depth data. What should I do for the case?

[yun-liu]

You should use the _rcf_pretrained_nyudimage.caffemodel which can be downloaded from this repository and the RCF-singlescale.ipynb for single-scale test, or RCF-multiscale.ipynb for multi-scale test.