Increasing sensitivity for locating solar panels

[JimWooten]

Is there a way to modify the parameters of the "Detect Objects Using Deep Learning" tool to increase the sensitivity of the tool to detect more solar panels using the Solar_Panel_Detection_USA dlpk model definition? The enclosed graphic indicates the location of detected panels using the tool (red rectangles) but it missed the large solar panel installation (blue rectangle). Thanks.

Jim

[PavanYadav]

@JimW  looks like the model is trained for solar panels on house roof tops. I will share your comment with the team that trained the model. 

[JimWooten]

Great, thanks Pavan.  A separate trained model for large solar panel installations would be very helpful.

[PriyankaTuteja · Accepted solution]

Hi @JimWooten,

We have two pretrained models available for detecting solar panels

1. Solar Panel Detecion-USA https://esriinc.maps.arcgis.com/home/item.html?id=c2508d72f2614104bfcfd5ccf1429284 

2. Solar Panel Detection-New Zealand https://esriinc.maps.arcgis.com/home/item.html?id=75b27dd904d34659bf6021689fa975e4

These models work well on the respective geographies mentioned un the name of the model. Could you tell us the geography where such solar panels are found? That said, if you like to detect solar panels in a different geography or of different size, you could fine-tune our models with your own data and obtain better results.

Additionally, if you would like to play around with parameters in the model arguments, you could lower the value of the threshold argument and see if the model detects the missed solar panels with lower confidence. 

Please let us know if you have any queries.

Thanks,

Priyanka