AI-BASED HELMET DETECTION PROJECT USING YOLOv11 AND STREAMLIT

AI-BASED HELMET DETECTION PROJECT USING YOLOv11 AND STREAMLIT 1Sams Tabrez, 2Keshav Kumar Mishra, 3Sachin Verma, 4Sachin Kumar, 5Komal Kusum, 6Rishav Raj, 7Himanshu Ranjan Das 1-6UG Student, Department of Electronics & Communication Engineering, Haldia Institute of Technology, Haldia, Purba Medinipur, West Bengal 7Department of Electronics & Communication Engineering, Haldia Institute of Technology, Haldia, Purba Medinipur, West Bengal 1tabrezmtp@gmail.com, 2Keshavkumarmishra35@gmail.com, 3sachinverma78760@gmail.com, 4sachin141ece@gmail.com, 5komalkusum18062005@gmail.com, 6rishavraj30420@gmail.com, 7das.himanshu69@gmail.com ABSTRACT In this paper we studied on how Wearing a safety helmet is one of the simplest ways to protect a worker's head at a construction site or factory, but checking this that all worker is wearing helmet by hand is very slow, tedious, and often uncertain. This paper presents an automatic helmet detection system built on the YOLOv11 deep learning model, trained on a labelled image dataset collected from the Roboflow platform. On unseen images, the model is showing precision of 77.7%, a recall of 90.1%, and a mean Average Precision of 91.3%, And for processing a single image it takes almost 11.8 milliseconds. The trained model was connected with web dashboard Streamlit, so that a safety officer with no technical background can upload a photo and instantly see whether the site is safe or not. The results confirm that how an AI model is used in human safety purpose.