Saathi

An AI powered agriculture utility platform

Features

• Plant disease detection(Detects upto 33 classes)
• Crop recommendation based on soil quality and environmental factors
• Information about different crops

Tech Stack

ML/DL: Tensorflow, Keras, Scikit

Web: Flask, Bootstrap

Installation

There are 2 parts in this project

ML/DL

In the models directory there are two folders 'recommender-models' & 'cnn'. Recommender-models has all the models related to crop recommendation system and cnn folder contains all the notebooks and models related to plant disease classification.

Web

  python3 -m venv venv
  cd Saathi/webapp
  pip3 install -r requirements.txt
  python3 setup.py

Results

Crop Recommendation

Algorithm	Accuracy	Precision	Recall	F1-Score
Logistic Regression	94.54	0.95	0.95	0.94
Decision Tree	97.72	0.98	0.98	0.98
SVM	9.09	0.59	0.09	0.11
Multilayer Perceptron	95.22	0.96	0.95	0.95
Random Forest	99.31	0.99	0.99	0.99

Plant Disease Detection

Architectures	Training Accuracy	Testing Accuracy	Validation Accuracy
VGG16	92.18	91.33	91.78
ResNet50	96.02	95.41	95.53
EfficientNetV2	96.06	95.53	95.83