The idea was to build a wildlife capture kit. It would use capture modules that use AI and machine learning on the edge to take photos of the wildlife species that you are interested in. When a photo of the species you are interested in is captured, a simple text message can be transmitted over the LoRaWAN network to tell what was captured. The key here is the trained machine learning model that can differentiate between species and only capture the ones you are interested in. The capture modules can be spread over vast distances in a mesh LoRaWAN network. Each module is capable of transmitting information over a line of sight distance of 10km.
The key part of this project was to train the model using Edge Impulse as this gives the untrained user the ability to get up and running with minimum effort. Throughout the project, I have reached out to @Eric Pan_Seeed as well as others on Discord trying to get the ability to train models with Edge Impulse, but as of now, this ability does not exist.
Unfortunately, I am not able to complete my project at this time. I do plan on finishing the project once the Edge Impulse model training is available.
Comments