rabiahdlearner / rice_yield_cnn

Rice Yield CNN is a model to estimate the rice yield based on RGB image of rice canopy at harvest. The model is developed based on more than 22,000 images and yield database collected across 7 countries.

Rice Yield CNN

Rice Yield CNN is a model to estimate the rice yield based on RGB image of rice canopy at harvest. The model is developed based on more than 22,000 images and yield database collected across 7 countries.
This project is the implementation of the paper "Deep learning-based estimation of rice yield using RGB image".

Performance

The model explained approximately 70% of variation in observed rice yield using the test dataset, and 50% of variation using the independent prediction dataset. The model is also able to forecast the rice yield approximately 10-20 days before harvest and is practically robust to the brightness, contrast or angle of the RGB image .

Conditions on estimation

RGB images that were captured vertically downwards over the rice canopy from a distance of 0.8 to 0.9 m using a digital camera should be input.

Environment on experiments

OS

• Ubuntu 18.04.5 LTS

CPU

• Intel(R) Xeon(R) W-2295 CPU @ 3.00GHz 18 cores

GPU

• NVIDIA GeForce RTX 3090 x2

CUDA

• Cuda compilation tools, release 11.3, V11.3.109

Python

• Python 3.8.8

Installation

1. Install depentencies.

pip install -r requirements.txt

2. Install Pytorch

Please install pytorch version compatible with your cuda version.

For example, If you use cuda version 11.3,

pip install torch==1.8.1+cu111 torchvision==0.9.1+cu111 -f https://download.pytorch.org/whl/torch_stable.html

3. Download pre-trained model from google drive.

mkdir checkpoints
wget "https://drive.google.com/u/0/uc?export=download&id=1XgTUGK8130gnY9AF3gYv9zhJSJaxhHVp" -O rice_yield_CNN.pth

Estimation

Run

python estimate.py --checkpoint_path checkpoints/rice_yield_CNN.pth --image_dir example --csv

You can find estimated yield on your console.

Below are meanings of options.

• checkpoint_path : Path to the checkpoint file you saved.

• image_dir : path to the directory where images are saved.

• csv: If you set this, csv of results will be generated.