Model description:

This model expects as input an image of a road, on which it detects potholes. 

​

Business case:

This model allows you to determine the number and size of potholes on a given section of road, so you can measure road quality. 

​

Number of classes: 1 (pothole)

 

Metrics:

detections_count = 952, unique_truth_count = 359
class_id = 0, name = Pothole, ap = 76.50%        (TP = 258, FP = 68)
for conf_thresh = 0.25, precision = 0.79, recall = 0.72, F1-score = 0.75
for conf_thresh = 0.25, TP = 258, FP = 68, FN = 101, average IoU = 61.88 %
IoU threshold = 50 %, used Area-Under-Curve for each unique Recall
mean average precision (mAP@0.50) = 0.764998, or 76.50 %

​

System requirements:

Inference time using CPU: 300 ms (on HP Laptop 15-DA0042NH (Processor: Intel(R) Core(TM) i7-8550U CPU))