This repository contains code to build the networks in "Deep Learning for Camera Autofocus" by Chengyu Wang, Qian Huang, Ming Cheng, Zhan Ma and David J. Brady.

(Last updated on: Dec 15, 2020)

• model_calibration.m

  Matlab code to calibrate the model given two images.

• estimator_discriminator.py

  Keras code to build the estimator and the discriminator in the AF pipeline.

  The images used in training can be found:

  CLIC dataset: http://www.compression.cc/2018/challenge/

  DIV2K dataset: https://data.vision.ee.ethz.ch/cvl/DIV2K/

• rl-agent

  Implementation of the reinforcement agent