#!/usr/bin/env bash
CUDA_VISIBLE_DEVICES=0 python main.py \
   --output_dir ./experiment_generator/ \
   --summary_dir ./experiment_generator/log/ \
   --mode train \
   --is_training True \
   --task SRResnet \
   --batch_size 4 \
   --flip True \
   --random_crop True \
   --crop_size 96 \
   --input_dir_LR ./train/train_global_reside/image_hazed/ \
   --input_dir_HR ./train/train_global_reside/image_real/ \
   --vgg_ckpt ./train/vgg_19.ckpt\
   --num_resblock 8 \
   --name_queue_capacity 4096 \
   --image_queue_capacity 4096 \
   --perceptual_mode MSE \
   --queue_thread 12 \
   --ratio 0.001 \
   --learning_rate 0.0001 \
   --decay_step 100000 \
   --decay_rate 0.1 \
   --stair True \
   --beta 0.9 \
   --max_iter 200000 \
   --save_freq 20000

#!/usr/bin/env bash
CUDA_VISIBLE_DEVICES=0 python main.py \
   --output_dir ./experiment_DHSGAN/ \
   --summary_dir ./experiment_DHSGAN/log/ \
   --mode train \
   --is_training True \
   --task SRGAN \
   --batch_size 4 \
   --flip True \
   --random_crop True \
   --crop_size 96 \
   --tmap_beta 2.0 \
   --input_dir_LR ./train/train_global_reside/image_hazed/ \
   --input_dir_HR ./train/train_global_reside/image_real/ \
   --vgg_ckpt ./train/vgg_19.ckpt\
   --num_resblock 8 \
   --perceptual_mode VGG54 \
   --ratio 0.001 \
   --learning_rate 0.00001 \
   --decay_step 100000 \
   --decay_rate 0.1 \
   --stair True \
   --beta 0.9 \
   --max_iter 200000 \
   --queue_thread 12 \
   --vgg_scaling 0.0061 \
   --pre_trained_model_type SRGAN \
   --pre_trained_model True \
   --checkpoint ./experiment_generator/model-170000

tensorboard --logdir [summary_dir]

#!/usr/bin/env bash
CUDA_VISIBLE_DEVICES=0 python main.py \
   --output_dir ./fog_exp4/ \
   --summary_dir ./log/ \
   --mode inference \
   --is_training False \
   --task SRGAN \
   --input_dir_LR ./test/image_hazed \
   --tmap_beta 2.0  \
   --num_resblock 16 \
   --perceptual_mode VGG54 \
   --pre_trained_model True \
   --checkpoint ./checkpoint/model-170000

python webapp.py --model_path {downloaded_model_folder}/model-170000

@InProceedings{10.1007/978-3-030-20873-8_38,
author="Malav, Ramavtar
and Kim, Ayoung
and Sahoo, Soumya Ranjan
and Pandey, Gaurav",
editor="Jawahar, C.V.
and Li, Hongdong
and Mori, Greg
and Schindler, Konrad",
title="DHSGAN: An End to End Dehazing Network for Fog and Smoke",
booktitle="Computer Vision -- ACCV 2018",
year="2019",
publisher="Springer International Publishing",
address="Cham",
pages="593--608",
abstract="In this paper we propose a novel end-to-end convolution dehazing architecture, called De-Haze and Smoke GAN (DHSGAN). The model is trained under a generative adversarial network framework to effectively learn the underlying distribution of clean images for the generation of realistic haze-free images. We train the model on a dataset that is synthesized to include image degradation scenarios from varied conditions of fog, haze, and smoke in both indoor and outdoor settings. Experimental results on both synthetic and natural degraded images demonstrate that our method shows significant robustness over different haze conditions in comparison to the state-of-the-art methods. A group of studies are conducted to evaluate the effectiveness of each module of the proposed method.",
isbn="978-3-030-20873-8"
}