Train RT-DETR on a Custom Dataset notebook seems that it doesn't work

Community Help
1

Hi I am using the notebook from this tutorial : How to Train RT-DETR on a Custom Dataset with Transformers in my Google Colab with my custom dataset and an error appears when running the following: trainer.train()

This is the error:

ValueError Traceback (most recent call last)
in <cell line: 11>()
9 )
10
—> 11 trainer.train()

8 frames
in collect_targets(self, targets, image_sizes)
27 post_processed_targets =
28 for target_batch, image_size_batch in zip(targets, image_sizes):
—> 29 for target, (height, width) in zip(target_batch, image_size_batch):
30 boxes = target[“boxes”]
31 boxes = sv.xcycwh_to_xyxy(boxes)

ValueError: not enough values to unpack (expected 2, got 1)

Thank you!

(Sorry i had change the code, now it is the correct error)

2

Hey! Could you send over your Google Colab Notebook as well as your dataset?
It looks like your there is only one of height/width in your image_size_batch. Would love to check out the Colab to look more into it though

3

Hello, the notebook is exactly the same as the one uploaded in the tutorial (this one: Google Colab), except for the cell where the dataset is uploaded, which I replaced with the download code for mine. I have my dataset on Roboflow; how can I share it with you?

4

Hi! Could you give me the link to your dataset? I’ll probably be able to access it on Roboflow Universe.

Sign in to Roboflow - would be this part: nathan-yan/train-window

5

Does this link work for you? Sign in to Roboflow
It’s the latest version, version 6.

6

This topic was automatically closed 21 days after the last reply. New replies are no longer allowed.