Running the model on a single image about egonet HOT 5 CLOSED

Hey Nicholasli1995,

Thanks for putting together this awesome repo, I really appreciate how thoroughly documented the setup is.

Would it be possible to get a few hints as to how to run the model on a single image as opposed to a directory?

I see that the data_loader in 'inference' is nearly directly from PyTorch, so it's probably my lack of experience with PyTorch that has me confused.

Separately, I also tried making my own directory and referencing it in my .yaml file, the following command still gave me an error about referencing the KITTI dataset

python .\inference.py --cfg "../configs/single_inference_test.yml" --visualize True --batch_to_show 1

I was a little bit surprised to see "training" in the file not found path as I'm not asking to train -- I would expect to have to place an analogous 'test.txt' in this directory with the single test image filename, but I don't think that's currently the issue.

Thanks! Clayton

Hi Clayton, can you explain your use case that "run the model on a single image as opposed to a directory"? Do you mean you don't want to initialize the kitti directory and simply process a single image? In that case, you can modify the image list .txt file so that it only contains the name of the image you need.

The error happened because the code assumes a directory structure as detailed in https://github.com/Nicholasli1995/EgoNet/blob/master/docs/preparation.md

from egonet.

Hi Nicholasli1995,

The intentional was to setup a pipeline so some other code could pass EgoNet an image without loading from a file but I did not fully grasp the D4LCN dependency earlier, so I think I can figure it out from here.

Thanks!
Clayton

from egonet.

Hi Nicholasli1995,

The intentional was to setup a pipeline so some other code could pass EgoNet an image without loading from a file but I did not fully grasp the D4LCN dependency earlier, so I think I can figure it out from here.

Thanks! Clayton

Sure. As long as there is a provided 2D region-of-interest EgoNet can be used. Any detector can be used apart from D4LCN to provide such proposals.

from egonet.

Just want to make sure, but EgoNet requires more than a 2D region of interest, right? For example, a YOLO model couldn't be used because it doesn't provide the y-rotation, 3d size, or 3d location?

from egonet.

Just want to make sure, but EgoNet requires more than a 2D region of interest, right? For example, a YOLO model couldn't be used because it doesn't provide the y-rotation, 3d size, or 3d location?

Not exactly. A user can also choose to feed 2D boxes to define the RoIs. The 3D inputs were just used to compute 2D RoIs by projecting them to the image plane. The "Reproduce results on the test split" section in https://github.com/Nicholasli1995/EgoNet/blob/master/docs/inference.md#reproduce-results-on-the-test-split shows the example of using only 2D boxes as inputs.

from egonet.