I am an agronomy graduate student looking to classify crops from weeds using convolutional neural networks (CNNs).
The basic idea that I am wanting to get into involves separating crops from weeds from aerial imagery (either captured by drones or piloted aircraft). The idea of the project that I am proposing involves spending some time driving around to different fields and capturing many images of both crops and weeds. These images will then be used to train a CNN that will classify aerial imagery on the location of crops and weeds. After classifying the imagery, a herbicide application map will be generated for site-specific weed control. This involves the integration of CNN classification and GIS technology.
My question is this: If you have an orthomosaic image generated from a drone, will images captured from a digital camera on the ground be effective for training a CNN that will classify high-resolution aerial imagery?
Being new to CNNs, I just didn't know if I had to use aerial imagery to train a CNN to classify aerial imagery, or if a digital camera will work just fine.