A confusing problem about how to use boxLabelDatastore correctly

I don't know how to deal with this issue and have tried to modify it several times, so I'm requesting someone who sees it to teach me.

thank you so much ！！！

[Jaimin]

Hi @ELZ,

Could you provide example data to help replicate the issue?

[ELZ]

Thank you for your reply, thank you very much.

I'm trying to train my annotated images with MATLAB's built-in Faster R CNN model, without using a pre-trained model.

This code was completed with the help of ChatGPT, but the problem has been solved.

However, in the following run, new problems arose, such as image size and input parameters.

Do you understand this? Could you please offer me some guidance? Please forgive me as I'm still in the process of learning.

Anyway, thank you very much for your attention.

Here are some relevant attachments

[Jaimin]

Could you please elaborate further on the issue?

[ELZ]

I'm glad you've been paying continuous attention. I've annotated some of the 800 pictures and stored the corresponding annotation data as txt files named after the pictures. After that, I planned to use Faster R-CNN and asked ChatGPT to help me generate an example. The problem now is that this example can't run. Maybe I have some misunderstandings about this issue. I currently have pictures (with different sizes) and annotation data. For the next step, should I first understand this model or consider switching to other models? I'm really sorry. I'm still learning. Thank you very much!

Answer 1

[Jaimin]

Hi @ELZ

Based on the described issue, I assume you are having trouble executing the deep learning model. After analysing the code provided in the file, I suggest using the “combine” function to merge the “imageDS_train” and “boxLabelDS_train” datastores.

Additionally, I recommend using a pre-trained model for your implementation, as it will reduce your training time.

For more information kindly refer to the following MathWorks documentation.

combine: www.mathworks.com/help/bioinfo/ref/bioread.combine.html

resnet example: https://www.mathworks.com/help/vision/object-detection.html

I hope this will be helpful.

[ELZ]

Thank you for your guidance, thank you very much.

I will try again

thank you so much