How to put my object detection dataset in this specific COCO annotation format?

[allansdefreitas]

Hello,

I'm running this Keras implementation of retinanet and it works perfectly on COCO dataset obtained from tfds.load().

I noticed that the downloaded TFrecords (from tfds.load() ) contain COCO annotations in a different way from traditional COCO annotations and that is the cause of all problems.

The COCO annotation format, creates one annotation element for each element in image, whereas Tensorflow COCO format (obtained from tfds), crates one single annotation for each image and inside this annotation, there is a list of objects and their positions (bboxes) That is the main difference.

Is there a way to put my dataset in this Tensorflow COCO format?

Thanks in advance,
Allan Freitas

System informations:
Ubuntu
Tensorflow: 2.9.1
Python: 3.7.13

[tilakrayal]

@allansdefreitas,
To expedite the trouble-shooting process, could you please provide the complete code to reproduce the issue reported here or the error log. It helps to debug the issue. Thank you!

[google-ml-butler]

This issue has been automatically marked as stale because it has no recent activity. It will be closed if no further activity occurs. Thank you.

[allansdefreitas]

Let me prepare and send to you...

[allansdefreitas]

@allansdefreitas, To expedite the trouble-shooting process, could you please provide the complete code to reproduce the issue reported here or the error log. It helps to debug the issue. Thank you!

@tilakrayal here you can find the notebook: https://github.com/allansdefreitas/Deep-Learning-Experiments/blob/master/retinanet_coco_error.ipynb

I do not know how to put my dataset in the tfds.load TFRecord internal format

[tilakrayal]

@allansdefreitas,
I was facing different error while trying to execute the given code. Kindly find the gist of it here. Thank you!

[allansdefreitas]

@allansdefreitas, I was facing different error while trying to execute the given code. Kindly find the gist of it here. Thank you!

It seems like if the content of TFRecords is encoded..

[tilakrayal]

@allansdefreitas,
Could you please provide the TFRecord to debug the issue. Thank you!

[allansdefreitas]

@allansdefreitas, Could you please provide the TFRecord to debug the issue. Thank you!

Hi, @tilakrayal.

You can download the TFRecords, running via terminal: curl -L "https://app.roboflow.com/ds/XqoiwGXC8L?key=NyDE34g8vI" > roboflow.zip; unzip roboflow.zip; rm roboflow.zip

[tilakrayal]

@allansdefreitas,
After uploading the tfrecord, I tried to execute the code and faced the same issue. Kindly find the gist here and please try to define the test_dataset . Thank you!

[google-ml-butler]

This issue has been automatically marked as stale because it has no recent activity. It will be closed if no further activity occurs. Thank you.

[google-ml-butler]

Closing as stale. Please reopen if you'd like to work on this further.

[google-ml-butler]

Are you satisfied with the resolution of your issue?
Yes
No