Pavement Crack Detection Deep Learning Module Not Picking Up Obvious Cracks
Hello, I'm new to using Deep Learning tools in ArcGIS Pro. The model I used was "Pavement Crack Detection".
I used drone imagery with less than 2cm resolution. I tried different iterations, like zooming in closer, setting a threshold, and setting some different environments. All of them returned the same lines. Is there a reason that it isn't picking up the areas circled in red below?
Thank you!
I am also experiencing issues with the tool failing to detect cracks in our dataset. The imagery was captured using an Autel Evo II Pro drone at 150 ft and 400 ft AGL, producing TIFF images mosaiced in PIX4D, with a native cell size of approximately 0.01384 meters in both the X and Y directions. We understand that the DLP was trained on imagery with a 2 cm per pixel resolution (0.02 meters). To align with the model's expected scale, we attempted to resample our TIFFs to 0.02 meters; however, the results were the same, and the tool does not detect most cracks after resampling. I did go through all the steps of installing the Deep learning Libraries. Has anyone found a solution?
