Rice Disease Detection

This was the first medium-scale capstone project of my team-mates and mine. In this project, the dataset was taken from Kaggle. It contained about 3000 images.

Dataset:

As stated above the total images that my team and me found was 3000. There are data of 5 diseased leaves and 1 healthy rice leaf.

Dataset Pre-processing:

I have increased the dataset using an image generator and enhanced the amount of dataset from 3000 to 8000. The techniques are:

• Rotation: 20%
• Zoom: 15%
• Width Shifting: 20%
• Height Shifting: 20%

All the data augmentation was performed using Keras's ImageDataGenerator class. Later on, normalisation and image segmentation was done on them after contour detection with edge detection.

Model and Deployment:

After pre-processing, 3 custom CNNs, 2 Resnet32(one with Adam optimiser and another one with RADAM optimise)s were used to predict diseases/healthiness from that dataset. The accuracy was about 83% on custom CNNs and 85% on Resnets. The model was used as a worker using MQTT data/message transferring broker service via an Android application.

Demo:

Full video link: (https://youtu.be/ePItve9IHrs)

Support or Contact

Md. Mahmudul Haque: mahmudulhaquearfan@gmail.com