Jaap de Ruyter van Steveninck

Hi @DorianBattivelli! Did you check whether this folder contains a .mat file? `/g/gross/Dorian2/Encoding-D.Battivelli-2025-11-10/training-datasets/iteration-0/UnaugmentedDataSet_EncodingNov10/`. Indeed, given by your error log, the '.' in your project title is likely the problem. Normally...

@AlexEMG, yes sounds good. The tensorflow modelzoo api was introduced in v2.3.8 (and the main branch seems to have been missing the `__init__.py` from the beginning). **Edit:** I have tested...

Hi @pvarsharavi, I don't know if you are still encountering problems with training. As mentioned by @maximpavliv, the performance is also highly dependent on your data quality. Besides looking at...

Hi @kross086, Not sure if you are still looking for a solution, but just wanted to check: did you try the following exact steps in Google Colab? 1. Start with...

Spotted and fixed some minor problems: - a circular import in `deeplabcut.post_processing` (only a problem when you load it individually). - `superanimal_name` argument no longer supported in `create_training_dataset`, which causes...

Also added a minor change to `deeplabcut.create_project.new`, to make sure that only video files and no directories are added to the final list of videos. (The `create_project` function unintendedly treated...

Hi @drhochbaum, unfortunately MPS support in PyTorch is still evolving and not yet on par with CUDA. This holds especially for detectors like Faster R-CNN which rely on complex ops....

@Jiefei-Wang, glad the quick fix worked for you. It was an issue with one of the packages that we are using, and we are currently implementing a more permanent fix...

@tlancaster6, thanks for your contribution, looks like a useful addition! I have two minor suggestions: 1. Instead of the fixed threshold of 0.25, it would be good to have an...

Hi @Alexmmvs, sorry for the late response. Are you still encountering this issue with DLC 3.0.0rc13? Would you perhaps be willing to share your data? In case you are willing...