Towards a novel wearable solution for citrus inspection using Edge AI

Wearable devices have become essential in our daily activities. Due to resource constraints, some devices cannot perform AI tasks, so they send those tasks to the cloud. However, these devices can face communication problems due to availability or latency. Therefore, Edge-computing-based intelligence is an alternative to overcome this issue. This article proposes a wearable device solution to aid citrus growers in oranges inspection. To validate the proposal, we profiled the prototype in two different scenarios with the integrated AI algorithm and compared it to two other platforms: The Sipeed M1 and The Raspberry Pi Zero W. The results indicated an improvement using a distributed architecture in the orange image classification task. This perspective allows for the preservation of the device's processing and power resources, thus possibly executing more complex tasks.

Source: https://www.researchgate.net/publication/362626607_Towards_a_novel_wearable_solution_for_citrus_inspection_using_Edge_AI
Dataset: https://www.kaggle.com/datasets/jonathansilva2020/orange-diseases-dataset
Code: https://github.com/JonathanCristovao/Multi-class-image-classification-mobilenet