FID computation in Jax/Flax

This is a port of mseitzer/pytorch-fid, which is a port of the original FID implementation (bioinf-jku/TTUR).

The parameters for the InceptionV3 network are taken from mseitzer/pytorch-fid. The FID scores are almost identical (absolute difference around 1e-7).
The only difference is that mseitzer/pytorch-fid resizes the images to 299x299 by default. In this implementation, the images are not resized by default. You can resize the images using the --img_size argument.

Installation

You will need Python 3.7 or later.

1. For GPU usage, follow the Jax installation with CUDA.
2. Then install:

   > pip install jax-fid
   

For CPU-only you can skip step 1.

Usage

Compute FID score

> CUDA_VISIBLE_DEVICES=N python -m jax_fid --path1 /path/to/dataset1 --path2 /path/to/dataset2

where N is the GPU index.

Pre-compute statistics for image directory

> CUDA_VISIBLE_DEVICES=N python -m jax_fid --precompute --img_dir /path/to/dataset --out_dir /path/to/stats

Arguments

--path1 - Path to image directory or .npz file containing pre-computed statistics.
--path2 - Path to image directory or .npz file containing pre-computed statistics.
--batch_size - Batch size per device for computing the Inception activations.
--img_size - Resize images to this size. The format is (height, width).
--precompute - If True, pre-compute statistics for given image directory.
--img_dir - Path to image directory for pre-computing statistics.
--out_dir - Path where pre-computed statistics are stored.

License

Apache-2.0 License