Pin Bolt State Identification Using Cascaded Object Detection Networks

Unmanned aerial vehicle-based transmission line inspections produce a large number of photos; significant manpower and time are required to inspect the abnormalities faults in such photos. As such, there has been increasing interest use computer vision algorithms automate detection defects these One most challenging problems this field is identification small pin bolts. In paper, we propose state framework cascaded by two object detectors. First, bolts located photos an initial detector. These expanded original picture cropped. processed then passed second detector that identifies three states pins: normal, missing, falling off. The proposed can attain 54.3 mAP 63.4 mAR our test dataset.