Peterande
provide me a pretrained model which is compatible for Yolov8 model with the object365
Please provide me a model that has to be compatible with the yolov8 architecture with the object365 classes.
I tried the above models but it has only the coco classes and not the object365
Have you double checked your statement and question, haven't you?
Pretrained weights on object365 are already avail, compatibility with another model doesn't make sense you know that right?
Have you double checked your statement and question, haven't you?
Pretrained weights on object365 are already avail, compatibility with another model doesn't make sense you know that right?
The pretrained model which was provided is a ".pth" file but usually for model prediction we use the ".pt" file. The pretrained model file's architecture is not mentioned, like which version of YOLO is used for the model.
Please provide me with some reference code on how to use the pretrained checkpoint model
@AnishMathewOommen you should first understand that this is like asking, "Give me a Mercedes for driving that is compatible with Ford !". You should first understand that D-FINE is a completely new model (with new architecture), different from YOLO.
Please provide me with some reference code on how to use the pretrained checkpoint model
Have you gone through this part?
@prashant-dn Thank you for the response. But for training we need to download the dataset. The size of the dataset comes nearly 715GB and I don't think it is possible for me to download this big dataset. Are you having a different approach for training? Did you download the entire dataset? Need your assistance
@AnishMathewOommen Why do you want to train the model on obj365 data? The pre-trained weights suffice there. You are probably looking for fine-tuning these models on your custom dataset or using this repo for training a model from scratch.
@prashant-dn I want to directly use the pretrained weights but it is in ".pth" format I am not sure how to use the pretrained weight.Could you please help me on how to directly use the pretrained model and then predict the detections present in the image?
It is specified in the README but here is a snippet: Once you have downloaded the pre-trained weight of your choice from the links mentioned in the README,
i=3_906_52_2024-12-08-17-25-14_1800_OG_0148.jpg // test image
w=PATH/TO/DOWNLOADED_MODEL.pth
export model=l
python3 tools/inference/torch_inf.py -c configs/dfine/custom/dfine_hgnetv2_${model}_custom.yml -r $w --input $i --device cuda:0