Water contamination from floating trash has become a serious environmental issue, requiring automated and effective waste collection systems. This work discusses the development of an IoT-based trash collector and an AI-based automatic water-floating robot for trash detection and removal. The trash collector uses a conveyor belt system to gather the floating garbage. The automatic water-floating robot uses a convolutional neural network (CNN) to recognize floating objects. The system has a conveyor mechanism for collecting trash, an Arduino Uno for controlling motion, and a Raspberry Pi for analyzing images. The robot uses a camera module to detect things while navigating aquatic bodies on its own. The robot stops, turns on the conveyor belt, and gathers the garbage if it detects an obstruction as trash. The robot cam avoids the thing if it is not identified as garbage. By minimizing manual intervention, the system improves the efficiency of water-cleaning processes. The proposed object detection system uses a dataset of 2,500 images, achieving 95% detection accuracy with a CNN model.