Introduction and my contribution
I am a co-author of the research paper titled "Image Acquisition and Deep Convolutional Neural Network Incorporated Cost-Effective Automated Nursery Disease Detection Method for the Tea Industry." My primary contribution to this work involved preparing the dataset and training the AI model using YOLOv7 and YOLOv8. I utilized Roboflow for data augmentation, starting with over 1,000 images related to blister blight detection. After data augmentation, we expanded the dataset to nearly 4,000 images, which were used for training the model.
Modeling system using solid works
The prototype system includes a method for capturing images while moving a camera above the trees. This approach allows the system to identify and activate the Fungicides to prevent the blister blight fungus from spreading. A SolidWorks model was developed for this purpose.
Final phototype system
The overall system consists of a camera for capturing images as video feedback, an AI model running on a computer, and an Arduino microcontroller that controls stepper motors based on the model's output. The system is designed to apply fungicides to the necessary plants. The Arduino is connected to the Python script via the PySerial library, enabling data exchange through serial communication.
Used Technology's
Torch
Ultralytics
google Colab
OpenCV
Final Working Video
