Alan Bi

Do any of the suggestions in #91 and #36 work? This issue occasionally comes up for some people and I'm not sure the best solution besides maybe double-checking/recreating environments to...

There shouldn't be much of a difference between image vs. video inference, the underlying model is still the same. Maybe try diversifying your set of training images to include more...

You could try using `core.Model(['foo'], pretrained=False)` and the loading the weights manually: ```python model.get_internal_model().load_state_dict(torch.load(path, map_location=model._device)) ``` Setting `pretrained` to false will prevent it from downloading a model pre-trained on COCO...

Could you take a look at #56 and let me know if that helps fix the issue?

The error in the image you shared seems to suggest there's a box with a width of 0, which would be invalid. If you modify your script to only print...

Looks like I made a mistake in my script, my bad - try this instead: ```python for i in range(len(dataset)): image, target = dataset[i] boxes = target['boxes'] for box in...

Could you take a look at #36 and see if any of those solutions work for you? Looking at the second image my guess is that it's the NaN issue...

> Hello, looked at #36 . Went through the code that generates the dataset again to make sure there were no issues with formatting. > Have created a new notebook...

It looks like your validation dataset doesn't have any images in it on line 548 - what's the output when you run `len(test_dataset)`? If it's 0, then this likely indicates...

Could you try some of the solutions listed in [this post](https://stackoverflow.com/questions/59129812/how-to-avoid-cuda-out-of-memory-in-pytorch) to see if any of those help?