An Efficient Machine and Deep Learning Techniques for Enhanced Crop Quality and Weed Control in Agriculture

Dr.B.Purushotham1, Gujjula Soni2, Devineni Dheeraj Chowdary3, Korrapati Madhav4,Rayadurgam Geyani5
1 Head of Department, Dept of Information Technology, SV College of Engineering, Tirupathi, India
2B. Tech, Dept of Information Technology, SV College of Engineering, Tirupathi, India.
3B. Tech, Dept of Information Technology, SV College of Engineering, Tirupathi, India.
4B. Tech, Dept of Information Technology, SV College of Engineering, Tirupathi, India.
5B. Tech, Dept of Information Technology, SV College of Engineering. Tirupathi, India.
Email: 1purushothamb@svce.edu.in, 2sonigujjula2004@gmail.com, 3dheerajchowdary10@gmail.com,
4Madhavkorrapati08@gmail.com, 5geyanir29@gmail.com

Abstract- Weeds infestation has direct consequences on agricultural yield and increases the usage of herbicides and labor cost. Conventional weed detection relies on manual inspection or classical machine learning techniques that require hand-designed feature extraction and do not work efficiently in changing field environment. While deep learning classification models like VGG16 and ResNet show an improvement in detection accuracy, they only perform image-level classification which fails to provide precise localization important for real-world applications. In order to eliminate these limitations, this work develops a real-time weed detection system based on the new YOLOv8 object detection architecture that achieves accurate localization in terms of bounding boxes and confidence-based classification while ensuring high speed.The model is trained on a labeled agricultural dataset with  data augmentation techniques which helps to enhance the generalization of model across different environmental conditions. We deploy the system through a web interface built with Streamlit that allows for real-time image uploads, video analysis, and control of confidence thresholds. Experimental results demonstrate reliable detection performance, highlighting the effectiveness of YOLOv8 for scalable, efficient, and AI-driven precision agriculture applications. Keywords: Precision Agriculture, Weed Detection, YOLOv8, Deep Learning, Real-Time Object Detection