Automated waste recognition multi-task model has been developed in response to the growing negative effects of inefficient waste processing and its advantages as it enhances performance and aligns more closely with how humans learn. This paper presents a clever method to enhance recycling procedures and reduce human sorting by utilizing deep learning and real-time object detection. The convolutional neural network (CNN) multi-task model was developed in the YOLO11 architecture: the brand-task recognizes items such as Pepsi, 7up, Chipsy-Kbab, Chipsy-Cheese, and the material-task recognizes waste into six categories: plastic, metal, paper, glass, cardboard, and trash. The material task determines the material category, while the brand task refines the recognition by brand. Using YOLO-based architecture for real-time recognition, multi-task model is trained on dataset of actual waste images. A smart recycling bin and a smartphone app that monitors recycling behavior and gives users prizes are also part of the smart waste collection system. The multi-task model’s potential for scalable, intelligent waste management and encouraging environmental responsibility is highlighted by the experimental results, which show that the material recognition task achieves with a final mAP@50 of 0.841 and the optimal F1- confidence of 0.78, the brand recognition task achieves mAP@50 of 0.904 and the optimal F1-confidence of 0.90, and the mAP@50 of the multi-task model is 0.87 and the optimal F1-confidence is 0.845.