Deeplab v3 segmentation model converted from TensorFlow to CoreML for on-device object segmentation. Code for the model and pretrained weights were taken from https://github.com/sthalles/deeplab_v3. Also, this notebook provides a good example of how to convert tensorflow model to CoreML and then test it.
- Download model checkpoint from here https://www.dropbox.com/sh/s7sx69pqjhrk0s4/AACXWCRd9JJ0zvcvDES9G3sba?dl=0.
- Just run notebook and then you will have .coreml model and tensorflow frozen graph.
Image is taken from https://instagram.com/asyaze.
I hope that I will add more interesting models and examples to this repo. If you want to add more examples of convertion or models, feel free to do it!