Robin Cole
Robin Cole
Provides interactive histograms and image plots to explore properties of the train/val data such as the size/area of bounding boxes. Contributing will require learning Panel. This might be overkill for...
blocked by https://github.com/airctic/icevision/issues/547
- In ice-vision notebooks we are applying default augmentations with `aug_tfms`
Noticed that roboflow actually stretches to 416x416 as a preprocessing step, so this might not be an issue
Hi @TheGreyDiamond this is not possible with my current data set as it does not include 'not a fire' images. However you could choose one of the listed data sets...
The below recipe appears fine: ```yml name: pytorch dependencies: - python=3.9 - pip - pip: - ipywidgets - conda - conda: - ipykernel - pytorch - torchvision ```
Hi @ioanfr this is not an issue as such, rather a feature request, for a validated env - note I did not test GPU access etc
An issue I encountered with this env: `AttributeError: module 'torch._C' has no attribute '_cuda_setDevice'` covered in https://github.com/pytorch/pytorch/issues/33655
I didn't create one, but if that's the cause of the error would be good to catch that