Open jessicametzger opened 3 years ago
Hello. As is suggested in this issue https://github.com/pierluigiferrari/ssd_keras/issues/359, the problem comes from the generator that yields a list instead of a tuple. You should be able to cast ret
to a tuple in this line
https://github.com/IntranelConsulting/ssd_keras_tf2/blob/183d57b1989fc6d732bf1028b6323bdd5771e78a/data_generator/object_detection_2d_data_generator.py#L1174
I am using ssd_keras with tensorflow2 backend to train the ssd300 model (using this ssd_keras_tf2 repository) and everything "works," except I keep running into this error when I start training:
They ran into the same thing when porting to tf2 in ssd_keras_tf2. Moreover, most parameters aren't changing, and the loss doesn't change after the first couple steps.
System Info
Full error message
Reproducible example
The error happens whenever I call
model.fit(...)
ormodel.fit_generator(...)
, where model is an ssd300 model, and where the backend is tf2. It happens whether I am using cpu or gpu. E.g. when I run the ssd300_training.ipynb tutorial, I get that error.I would really appreciate it if someone trying to use tf2 backend found a solution for this, because I'm pretty sure I'm not the only one.