The development of current technology makes easier for human beings in overcoming work. One of them is a selection of objects on industry. On a large-scale industry, the process of selection objects is done continuously, and it requires a high degree of accuracy. The work goes on for 24 hours unceasingly, which is hard for human. Therefore, it needs an instrument that works automatically to ease the selection process of objects. In this research, a prototype that separates objects based on shape is conducted. Shape recognition process utilizes canny edge detection method with microcontroller ATmega16 as a whole system controller. According to the testing result, the prototype can recognize and separate objects a circle, square, rectangular, and equilateral triangles. The ratio roundness (metric) values for circle, square, rectangular, and equilateral triangle are 1.00-0.90, 0.90-0.83, 0.81-0.70, and <0.70, respectively. Moreover, the level of success for separating circle object, square object, rectangular object, equilateral triangle object, and whole system are 100%, 80%, 100%, 100%, and 85%, severally.