Neural network on an ESP32Cam microcontroller for fruit and vegetable

Madiop DIOUF, Cheikh Sidy Mouhamed CISSE, Birahime DIOUF, Ibra DIOUM, Idy DIOP

Computer Vision (CV), Artificial Intelligence (AI), Machine Learning (ML), Deep Learning (DL).

Thanks to machine learning (ML), particularly deep learning, agriculture is undergoing a profound transformation. Convolutional neural networks (CNN), particularly suited to image analysis, have made it possible to develop computer vision systems capable of accurately identifying a wide variety of fruits and vegetables. Our project explores this avenue by using an ESP32-CAM module to create a deep learning-based fruit recognition system. The goal is to automate harvesting by leveraging the ability of CNNs to extract complex visual features, such as fruit shape, color and texture.

We used the Edge Impulse platform to train our model and deployed it on the ESP32cam module, the model results are displayed as output on the serial monitor with 98% accuracy orange and 87% banana. This work aims to provide a solid foundation for future research exploring the application of deep learning (DL) to fruit detection and recognition in the context of automated harvesting.

https://dx.doi.org/10.31873/IJEAS.11.11.04