I’m using this keras based implementation of RetinaNet and I’ve found it really well designed and maintained but I’m now getting thousands of , many of which are false positives, and I’m curious if y’all have suggestions for or methods for reviewing the results from a CNN?

I’ve been using the VGG Image Annotator to create my labeled and I’m considering using that to go through review the detections too but I wanted to see what everyone suggests and if there are any rapid tools for reviewing the results from a net.

I’m also considering building something custom in to go through and mark detections as correct or incorrect but wanted to check landscape beforehand.



Source link
thanks you RSS link
( https://www.reddit.com/r//comments/95et15/d_tools_for__neural_net/)

LEAVE A REPLY

Please enter your comment!
Please enter your name here