The Elderly people’s freedom and health are seriously threatened by falls, which can often lead to serious injuries and a reduced quality of life. Falls are a major threat to elderly health and independence. Existing fall detection systems rely on wearables, which can be inconvenient. This proposal explores a camera-based system using Convolutional Neural Networks (CNNs). CNNs excel at recognizing patterns in visual data. Here, the system would continuously analyse real-time video to identify changes in posture, gait, and environmental hazards that signal fall risk. By learning from a large dataset, the CNN would predict falls well before they happen, unlike reactive systems that only detect falls after they occur. This proactive approach allows caregivers and emergency services to intervene sooner, potentially preventing falls or lessening injuries. This technology has the potential to improve senior safety and well-being, allowing them to live more independently with confidence.