Floating solid waste has become a significant threat to aquatic ecosystems, water quality, and public health. Traditional monitoring methods are often manual, time-consuming, and geographically limited, making them insufficient for real-time environmental surveillance. This study introduces an intelligent system that detects and monitors floating solid waste in water bodies using a real-time YOLOv12 deep-learning model. The proposed system combines drone video recording capabilities with mobile screen mirroring functions, on-device processing, and automated alert systems to develop an environmental monitoring solution that operates in the field and can be scaled up. The model demonstrated outstanding performance under challenging visual conditions, achieving a mean Average Precision (mAP) of 0.978 () and 0.847 (). The new architectural features of YOLOv12 enable these results through a combination of R-ELAN modules, Flash Attention, and EIoU loss. The detection pipeline operates on standard consumer hardware to transmit live aerial footage from mobile devices to laptops for real-time inferences. The system uses Telegram to send rapid notifications. The system reduces repeated alerts through redundancy filtering using the pollution coverage ratio to determine contamination severity. This system distinguishes itself from existing solutions because it merges drone video analysis with intelligent alerting and immediate spatial pollution assessment. The proposed method demonstrates both practicality and strength according to the results, which establishes a significant advancement toward better environmental monitoring systems.