bcv-uniandes / png Goto Github PK

Hello! Thank you for sharing the code. However, I can not download the Panoptic Narrative Grounding Benchmark from the project webpage due to connection time out. Could you please help me with this issue?

What is the DATASET_SUFFIX? I have seen how to generate or use DATASET_SUFFIX.

Thanks.

Hi,
This is a very interesting work. I cannot train the model because pre-computed features are not available for the training split. So whether you plan to release the training features?

Thanks!

Hello,

Thank you for providing the code, and congratulation on the paper acceptance, this is a very interesting task.

It is very helpful to provide the baseline model, however, I’d argue that it would facilitate adoption if a standalone reference evaluator for the task was provided. Something with minimal dependencies (eg no assumption on the DeepLearning framework used, although numpy/sklearn are probably ok), that takes in the path to the annotations, as well as the model predictions in a documented format (json or dict for eg), and spits out the official metrics. Some basic checks could also be carried out, eg that all predictions are present, that there is no duplicate, that the masks have the correct resolution,...

As far as I can tell, the current evaluation code is too tightly integrated with the model to be used independently, for example taking care of dealing with distributed aspect, relying on internal configuration classes, and more importantly, integrating the forward pass directly inside the evaluation function.

For examples of such standalone evaluators, I refer to the panoptic evaluation toolkit for coco: https://github.com/cocodataset/panopticapi or for a more closely related task, the referring expression segmentation evaluator on PhraseCut: https://github.com/ChenyunWu/PhraseCutDataset

Looking forward to working on this task,

Best regards,
Nicolas Carion

Hi, this is a nice work. The pre-computed features are empty on your website.
Looking forward to your reply.

Jiaheng.

In "panoptic_narrative_grounding.py", the definition of "mask_transform" is as follows:

,which takes the "PIL Image" as input.
However, the type of the input "mask" is a tensor, there is an error ["TypeError: img should be PIL Image. Got <class ‘torch.Tensor’>"] when I run your repo.
Therefore, I change this code in this way:
, and the code can run successfully.

A similar change is also in the 'train_net.py'

The types of "p" and "t" are both tensors.

Is it OK?

Thanks for your nice work and code.
Looking forward to your reply.
Jiaheng.

Hi, thank you for providing the code
I want to know how to get "panoptic_seg_predictions".