• 

  Project Workflow

• 

  Training Loss and Accuracy

Hoya_Plant_Disease_Predictor

Patravaidyam - diagnosis of plant leaf disease.

Goal:

The main goal is to predict the status of the plant i.e whether it is healthy or diseased then which disease is it and to provide the remedies for it.

Prerequisites:

1. Google Colab
2. Plant Disease Dataset
3. Python
4. Javascript
5. CSS
6. Html

Libraries:

1. Tensorflow
2. Keras
3. Numpy
4. Matplotlib

How the model works:

The user needs to insert input leaf image into the web application, using the Upload Image button and then the input image is sent to the classification model here we have used MobileNet architecture and the training is performed on the train and valid dataset(fetched from Kaggle), followed by testing and the predicted output is sent back to the web application, where we get the type of plant of input leaf image and followed by its health status and the remedy button gives you the prescription of the specific disease.

Challenges:

The dataset was very large it contained nearly 70,000 samples due to which the model was taking too much for the training. So we decided to reduce the samples and we took 4000 samples.

Achievements:

We have successfully completed the training for our with nearly 75% accuracy. We have also developed a web interface for our project.

What's Next:

We will try to increase the accuracy and also try to improve the web interface.

Built With

• css
• google-colab
• html
• javascript
• keras
• python
• tensorflowjs