Swimmers can find their lives at risk due to insufficient visibility and unforeseen obstacles in open waters. Drowning takes over 320,000 lives every year which makes it a global concern and necessitates the search for innovative ways for improving safety in open waters. The research proposes a real-time alert and direction guidance system for swimmers equipped with a deep learning model based on Convolutional Neural Network architecture called YOLOv10 for obstacle detection. The YOLO v10 model employed in the system runs on a Raspberry Pi 5 (RPi5) microcomputer and is trained with custom datasets containing images of commonly encountered open water obstacles like boats, jet skis, fishing nets, and marine creatures. The RPi5 system can be mounted on a drone with a high-resolution camera to enable autonomous tracking of swimmers and potential obstacles in their route by capturing and processing real-time video frames. Each of the processed video frames is passed to the trained YOLOv10 model to detect obstacles and the locations of threats. Wearable wristband devices incorporating ESP32-S3 microcontroller with a haptic motor provides haptic feedback to the swimmer wearing them suggesting safer swimming routes. If an object is detected on the swimmers’ left side, vibrational impulse is provided on the right hand wristband and vice versa letting the swimmer know the direction they should continue swimming. The prototype underwent testing in simulated marine environments where it provided accurate obstacle detection and real-time haptic feedback. This system improved the awareness of swimmers and thus lowered the chances of a collision with an obstacle. This novel proactive approach proved its potential in increasing safety during recreational and competitive swimming and open water training. This real-time obstacle detection and route guidance system can certainly reduce the chances of drowning threats and improve swimmer safety in turbulent aquatic scenarios. The prototype was successfully tested in real open-water conditions, with effective obstacle detection and real-time haptic feedback validated by positive swimmer feedback on system usability and comfort. Quantitative evaluation of detection accuracy and processing latency remains a part of future work.