talmo
commented
2 years ago Hi @om-git216,
From the error it looks like the model doesn't even exist (D:/CuSO4/Standardising arena size/Medium size arena\models\singleanimal_trial220113_151510.single_instance.n=20) -- this is probably because the training failed altogether.
We can find some more troubleshooting hints if you happen to still have the command line logs from the training (should be just before the logs you pasted).
Another quick tip: even though this is a single animal, it might be a bit easier to actually use a top-down multi-animal model. The top-down model will first detect the animal and then crop it before estimating the other landmarks. This will be helpful for your case because the fly is a lot smaller than the full size of the image. When your model is ready for tracking, we can then filter it down to a single instance to get rid of any extraneous detections.
owadhwa
jfrie
Hi,
I have been trying to use SLEAP to estimate the poses of Drosophila single fly videos. However, I get the following error message after starting to train the model:
The terminal says the following:
Traceback (most recent call last): File "C:\Users\Omika\anaconda3\envs\sleap\Scripts\sleap-track-script.py", line 33, in
sys.exit(load_entry_point('sleap==1.1.5', 'console_scripts', 'sleap-track')())
File "C:\Users\Omika\anaconda3\envs\sleap\lib\site-packages\sleap\nn\inference.py", line 3034, in main
predictor = _make_predictor_from_cli(args)Args:
File "C:\Users\Omika\anaconda3\envs\sleap\lib\site-packages\sleap\nn\inference.py", line 2969, in _make_predictor_from_cli progress_reporting=args.verbosity, File "C:\Users\Omika\anaconda3\envs\sleap\lib\site-packages\sleap\nn\inference.py", line 2676, in load_model batch_size=batch_size, File "C:\Users\Omika\anaconda3\envs\sleap\lib\site-packages\sleap\nn\inference.py", line 154, in from_model_paths batch_size=batch_size, File "C:\Users\Omika\anaconda3\envs\sleap\lib\site-packages\sleap\nn\inference.py", line 1170, in from_trained_models confmap_keras_model_path, compile=False File "C:\Users\Omika\anaconda3\envs\sleap\lib\site-packages\tensorflow\python\keras\saving\save.py", line 186, in load_model loader_impl.parse_saved_model(filepath) File "C:\Users\Omika\anaconda3\envs\sleap\lib\site-packages\tensorflow\python\saved_model\loader_impl.py", line 113, in parse_saved_model constants.SAVED_MODEL_FILENAME_PB)) OSError: SavedModel file does not exist at: D:/CuSO4/Standardising arena size/Medium size arena\models\singleanimal_trial220113_151510.single_instance.n=20\best_model.h5/{saved_model.pbtxt|saved_model.pb} data_path: D:/CuSO4/Standardising arena size/Medium size arena/medium size arena_camera at top_no zoom.mp4 models: ['D:/CuSO4/Standardising arena size/Medium size arena\models\singleanimal_trial220113_151510.single_instance.n=20'] frames: 1227-58541 (20) only_labeled_frames: False only_suggested_frames: False output: D:/CuSO4/Standardising arena size/Medium size arena\predictions\medium size arena_camera at top_no zoom.mp4.220113_154240.predictions.slp Process return code: 1
Could you please help me out resolve this issue?
Also, just for more information, the validation of the model is also quite off: (though I would expect this because the model is not fully trained yet)
Thanks in advance,
Omika Wadhwa