Harshitha Davuluri

phn : 9949283666

gmail: harshitha.davuluri219@gmail.com

Lumpy Skin Disease Detection Using Explainable AI (XAI)

Objective:
The project aimed to develop an AI-driven diagnostic framework for early and accurate detection of Lumpy Skin Disease (LSD) in cattle. By leveraging Convolutional Neural Networks (CNNs) and Explainable AI (XAI) techniques, the solution not only achieved high accuracy but also enhanced model interpretability to foster trust among end-users like veterinarians and livestock managers.

Key Features:

• AI Model: Designed a robust image classification system using the EfficientNetBO deep learning architecture, fine-tuned for detecting skin anomalies in cattle.
• Data Augmentation: Implemented advanced techniques such as rotation, flipping, zooming, and brightness adjustments to enhance model robustness and generalization across diverse conditions.
• Explainability: Integrated Grad-CAM (Gradient-weighted Class Activation Mapping) to provide visual insights into model predictions, ensuring transparency and reliability.
• High Performance: Achieved an impressive accuracy of 95.59% and high precision and recall metrics, making the system a reliable tool for veterinary diagnostics.

Impact:
The project addresses the limitations of traditional diagnostic methods, which are often time-consuming and inaccessible in rural settings. By automating disease detection, this framework offers a scalable, cost-effective, and real-time solution for livestock health management, aiding in the early containment of outbreaks.

Future Scope:

• Expansion to include detection of multiple livestock diseases.
• Optimization for real-time deployment using lightweight models like MobileNet.
• Integration with IoT devices for remote monitoring and diagnostics.

Automated Weed Cutter

• Developed and implemented code for an automated weed cutter using Arduino technology.
• Designed the system to effectively cut unwanted weeds and waste before and after farming cycles.
• Utilized sensors and automation techniques to enhance the precision and efficiency of the weed-cutting process.
• Successfully completed the project, demonstrating significant improvements in agricultural maintenance and productivity.
  
  Indoor Air Conditioning Monitor
• Developed an Arduino prototype to monitor indoor air quality and detect pollution levels in a room.
• Implemented sensors to measure various air quality parameters such as temperature, humidity, and particulate matter (PM2.5 and PM10).
• Programmed the Arduino microcontroller to process sensor data and display real-time air quality readings.
• Integrated an alert system to notify users when pollution levels exceed safe thresholds.
• Documented the development process and created user manuals for easy setup and operation.