Pest Classification and Detection System using Deep Learning

Pests significantly affect agricultural yields, leading to a decline in productivity and nutrient depletion. Excessive pesticide usage in the agricultural industry often results in increased pesticide residues, disrupting the food chain and causing adverse effects on human health and the environment. Traditional methods for detecting pests are limited, inefficient, and time-consuming, relying on manual identification of specific traits. These challenges have a substantial economic impact, with farmers experiencing financial losses and tragic loss of lives. Early detection of pests allows farmers to avoid premature pesticide application, promoting a more sustainable and environmentally friendly agricultural practice. This project introduces an innovative approach for crop pest detection using deep convolutional neural networks. The proposed model accurately identifies some common agricultural pests, providing an efficient means for classification based on photographs captured online or offline. Leveraging pre-trained models retrained with a more extensive and diverse dataset enhances accuracy, even when working with a smaller pest-specific dataset. This advancement in pest detection technology empowers farmers to make informed decisions and effectively manage pest-related challenges in agriculture, contributing to a more sustainable and productive sector.

PROPOSED SYSTEM

The existing farming information sites are not dealing with the pest classification and detection. We propose such a project proposal with an aim that whenever this comes to real world Pest Detection System is definitely going to meet the needs of common man in farming field to some extent. In our system, there are user to ask any queries directly to the agricultural officers or experts.

FEATURES OF PROPOSED SYSTEM

1. The user can ask any doubts regarding the pests.
2. The user gets the detailed information like what kind of pest, what pesticide are used to prevent.
3. The agricultural officer/expert can view the details of pests.

FUNCTIONS OF PROPOSED SYSTEM

1. Automated Pest Detection: Your project automates the pest detection and classification process using a deep learning model.
2. Efficient: It can process images rapidly, making it more efficient than manual inspection.
3. Accurate: The deep learning model has the potential to achieve high accuracy in identifying pests.
4. Scalable: It can be deployed on a larger scale, enabling the inspection of a greater number of images.
5. User-Friendly: User Interface (UI) makes it accessible to users without expertise in pest identification.
6. Detailed Information: Your project provides detailed information about detected pests, including descriptions and pesticides information.

Tools used 🛠️

1. Platform: Jupyter notebook, Google Collab
2. Libraries Used: Flask, Numpy, Pandas, Torch, PyMongo

Prediction page

Result page

Officer Dashboard

Steps to run ⚙️

1. First make folder models and place pest_model.pth into that folder.
2. Run the command

   python main.py