For face detection, when sampling, the image reconstructed from recons_image = self.model.decode_first_stage_with_grad(pred_x0) in ddim_with_grad.py produces recons_image which has values higher than 1 instead of [0,1]. I was wondering did you also face this problem?
Due to which mtcnn is unable to produce any valid bounding box and gives error
TypeError: cannot unpack non-iterable NoneType object
Also, I changed self.model.module.decode_first_stage_with_grad(pred_x0) to self.model.decode_first_stage_with_grad(pred_x0) as it was giving an error as model.modules doesn't exist.
For face detection, when sampling, the image reconstructed from
recons_image = self.model.decode_first_stage_with_grad(pred_x0) in ddim_with_grad.py
produces recons_image which has values higher than 1 instead of [0,1]. I was wondering did you also face this problem? Due to which mtcnn is unable to produce any valid bounding box and gives errorTypeError:
cannot unpack non-iterable NoneType objectAlso, I changed
self.model.module.decode_first_stage_with_grad(pred_x0) to self.model.decode_first_stage_with_grad(pred_x0)
as it was giving an error as model.modules doesn't exist.Thanks.