Power line recognition method via fully convolutional network

View Full Paper

Ask AI

Bookmark

Share

Key Points

Develop an automated deep learning framework using a modified fully convolutional network to accurately detect and segment power lines from optical and infrared images for low-altitude aviation safety.
Generated paired synthetic datasets featuring power lines against complex backgrounds with corresponding pixel-level annotations.
Adapted a U-Net architecture trained on synthetic data, incorporating focal loss to address severe foreground-background pixel imbalance.
Evaluated performance on an aerial inspection dataset comprising 4,000 infrared images and 4,000 visible-light images against baseline random forest models utilizing VGG16 features.
Achieved a power line recognition rate exceeding 99% alongside a false alarm rate below 2%.
Demonstrated superior pixel-level semantic segmentation of thin power line structures compared to conventional feature extraction and machine learning baselines.

Answers are grounded in this paper and cite their sources.

Validation study demonstrates accurate power line detection in aerial imagery using an adapted U-Net, suggesting enhanced low-altitude flight safety.

No abstract yet. The journal hasn't published one we can show.

No takes yet. Share an insight, caveat, or question.

Liu et al. (2020) studied this question.

View Full Paper

Ask AI

Bookmark

Share