Implementation of Agricultural Produce Segregation Using Image Processing Algorithm

Sorting of produce is an integral part in farming, agricultural and distribution sectors. Most farmers resort to hand picking to remove the contaminants which are mostly fruits or vegetables infested with some type of fungus, worms or insects. Hand picking the produce leads to too much wastage of time. It is also not perfect due to human error which results in end consumer receiving products with subpar quality leading to damage of brand name. This results in market loss resulting in reduction of profit. The proposed system aims to target this activity by sorting the produce based on its appearance. This is done by processing the image of the fruit or vegetable and classifying it as whether it is fresh or rotten. Capturing the image of the produce and processing it is done with Espressif ESP32 Cam Module. The dataset containing all the different class is processed and stored as a pre-defined dataset in online servers. This dataset is accessed by the microcontroller to classify the image of produce. RetinaNet algorithm is used to classify the image based on their appearance. Since the produce needs to be classified individually, a conveyor system is necessary. The ESP32 Cam Module is mounted on top of the conveyor setup. A lever is placed at the end of the conveyor to segregate the produce. The ESP32 Cam Module controls the lever based on the result from processing the image. Overall, the proposed system will drastically reduce the time consumed to segregate the produce while increasing the efficiency of sorting.