How to test a model on a custom dataset?

[David-Biggs]

Thanks for the amazing repo!!

I have fine-tuned a model using my custom train and Val datasets. I have the config file as well as the weights file of the model.
I would like to test the model on unseen images in my test dataset. Is it possible to use /tools/test.py to test a model on a custom dataset? If so how should I go about doing that?

Many thanks,
Dave

[hhaAndroid]

@David-Biggs sorry, currently not supported. If you just want to get the prediction result, you can refer to demo/image_demo.py

[David-Biggs]

Hi @hhaAndroid, thanks for your response.

So there is no way of testing models trained on a custom dataset (trained on train dataset, validated using val dataset and tested on an unseen test dataset )?

I was under the impression that testing models on unseen data is a very important aspect of model comparison.

I've trained 6 different models and I've fine-tuned them using my custom val dataset and now I would like to compare when using mAP on my custom test dataset. Is there a way to do this?

Many thanks