The increasing problem of underwater trash and its detrimental impact on marine ecosystems necessitates effective detection and mitigation strategies. This work presents an approach for underwater trash detection by integrating the YOLOv7 deep learning model with a Flask web application. The proposed system enables users to upload images or videos through the web application’s user interface for real-time detection of underwater trash objects. To train the YOLOv7 model, a comprehensive dataset of annotated underwater trash images is curated, encompassing diverse types of marine debris commonly encountered in aquatic environments. The model is fine-tuned using this dataset to accurately recognize and localize underwater trash objects in real-time. The Flask web application serves as a user-friendly platform, allowing individuals to easily upload images or videos from their devices for analysis. Once uploaded, the application processes the media content using the trained YOLOv7 model. It enables the monitoring of marine pollution, empowers users to identify underwater trash hotspots, facilitates cleanup initiatives, and promotes awareness about the significance of preserving marine ecosystems. The user-friendly nature of the web application encourages active user participation and engagement in combating underwater trash. The system has the potential to aid in the preservation of marine environments by facilitating proactive efforts to mitigate the impact of underwater trash.