harmluSICKAG
harmluSICKAG
@ArgoHA in the standard configs, multiscale is turned on until epoch 71. This regularization technique is often used in all sorts of model trainings. In yolox, e.g., input sizes during...
I will add this as a feature if that is ok with you.
See https://github.com/Peterande/D-FINE/pull/142
It is pretty straight forward to convert the model to openvino format ``` model = ov.convert_model( onnx_file, input=[("images", [1, 3, dim, dim]), ("orig_target_sizes", [1, 2])], ) ov.save_model(model, output_file, compress_to_fp16=args.fp16) ```
If not, you should shfit them to start with zero. ``` import json def shift_category_ids(coco_file_path, output_file_path): try: # Load the COCO JSON file with open(coco_file_path, 'r') as file: coco_data =...