Stasja

@normalclone were you able to understand where this issue came from? I have the same problem: one single image from demo returns different results on different runs

Actually, I managed to fix it! If anyone else has this error, the main problem is the transformation of the input image in ```mmfashion/utils/image.py```. **How to fix:** 1. find where...

@ebernalg92 ``` model.load_weights(model_path, by_name=True,exclude=[ "mrcnn_class_logits", "mrcnn_bbox_fc","mrcnn_bbox", "mrcnn_mask"]) ``` But actually, I also checked, and it changes the prediction from time to time (with the same weights, taken from the same...

@ebernalg92 that worked!! Thank you so much, now it's super precise! But if it's not too difficult could you, please, explain, why?

@hjm1990818 If you have the json files as annotations, it should be enough. You can check in the Shapes and the Balloon samples, there is a class Dataset with the...

@KnightOfTheMoonlight @walsvid In the ```rendering_metadata.txt``` there are 5 parameters. so, 1 - azimuth 2 - elevation 3 - tilt 4 - distance to the zero to the object 5 -...