RGB2NIR_Experimental

This repository contains several image-to-image translation models, whcih were tested for RGB to NIR image generation. The models are Pix2Pix, Pix2PixHD, CycleGAN and PointWise.

Dataset

The dataset including RGB and NIR images is located here: https://drive.google.com/drive/folders/1uJx_SLi0ePYqhn-lJ8p5whAy5spUsect?usp=sharing, where "RGB_Images" and "MSI_images containg RGB and mltispectral images acquired with digtial and multispectral 9 band camera at the same time periods, respectively. "RGB_NIR_Decayed_apples" contains RGB and multispectral images acquired with multispectral camera. The models were trained and tested on images from "RGB_Images" and "MSI_Images", respectively.

How to read and understand RGB image

20_12_26_22_15_00_Canon_top_all_on.jpg: 20 - year (it was 2020 for current image); 12 - month (December); 26 - date/day (26th); 22 - hours (22:00 or 10 pm), 15 - minutes, 00 - seconds; top_all_on - location.

How to read and understand multispectral image

set10_20201226_221732_686_00000_channel0.png: set10 - number of set; 2020 - year (it was 2020 for current image); 12 - month (December); 26 - date/day (26th); 22 - hours (22:00 or 10 pm), 17 - minutes, 32 - seconds; 686_00000_ - number of image; channel0 - number of multispectral camera's band.

The values of each multispectral camera's channels

channel0 = 561 nm, channel1 = 597 nm, channel2 = 635 nm, channel3 = 635 nm, channel4 = 724 nm, channel5 = 762 nm, channel6 = 802 nm, channel7 = 838 nm; channel8 (panchromatic) = 0 nm.

Citation

In this project we used the code and methodolgy by:

@inproceedings{CycleGAN2017, title={Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks}, author={Zhu, Jun-Yan and Park, Taesung and Isola, Phillip and Efros, Alexei A}, booktitle={Computer Vision (ICCV), 2017 IEEE International Conference on}, year={2017} }