Using Image Recognition and Monitoring Fruit in the Farmland Environments

In the paper, a smart embedded microcomputer system will be constructed and will recognize objects within images streamed in real time. After the manipulator has been trained to move automatically, it will be equipped to control objects without a human operator under various environments. To achieve this aim, the image recognition system should be able to identify objects as well as humans. Specifically, it must undergo neural network training, learn the features of target objects, and measure the relative distances between the objects in a space to determine their respective locations and patterns of movement.