如何使用自定义数据集运行 Keras Retinanet 对象检测代码示例？

Question

I'm trying to implement the Keras Retinanet example from the source:

https://keras.io/examples/vision/retinanet/

and I do get the example run perfectly with the COCO-dataset. Now I would like to run the example with my own custom object detection dataset. After doing couple of days some research on the web it still isn't that clear for me, how I would need to edit the example code to use my own dataset (that is a set of .jpg images + annotation .xml files produced with labelImg ).

I suspect I would need to get familiar with the tensorflow_dataset-library ? My guess is that this part in the code is where I should do the customization for my own dataset:

https://keras.io/examples/vision/retinanet/#load-the-coco2017-dataset-using-tensorflow-datasets

Any advices or good references worth investigating to get the Keras Retinanet object detection example working with my own dataset?

Answer 1

I found one nicely instructed collab link for the implementation of Retinanet: https://colab.research.google.com/github/benihime91/pytorch_retinanet/blob/master/demo.ipynb

Also there is another link: https://www.programmersought.com/article/73514120308/

Both of the above pasted links are well described and are meant for custom dataset.