Ground Truth of Power Line Dataset (Infrared-IR and Visible Light-VL)

Published: 10 February 2017| Version 1 | DOI: 10.17632/twxp8xccsw.1
Contributors:
Ömer Emre Yetgin,

Description

Despite a relatively crowded literature about detection of power lines for aircraft safety, the works are mostly optimized with very few training images; some even use artificially generated images. The IR imaging case is even less utilized. The reason is clearly the tremendous workload to obtain real images. With this demand in mind, the authors cooperated with the Turkish Electricity Transmission Company (TEIAS) to obtain video captures from actual aircrafts. Later, the authors made a thorough inspection over the video frames to isolate, capture and clean thousands of valuable images. In this content, 50 IR and 50 VL images are acquired and scaled to a size of 512x512. The IR folder contains IR images with power line, ground truths and overlay images of these images. The VL folder contains VL images with power line, ground truths and overlay images of these images. The IR and VL groups were deliberately constructed to contain both regular and especially confusing scenes. The videos were captured from 21 different regions all over Turkey at different seasonal days. Due to varying background behavior, varying temperatures and weather conditions, and varying lighting conditions, the achieved positive set contains several difficult scenes where low contrast causes close-to invisibility for power lines. The original video resolutions were 576x325 for IR and full HD for VL, however, the captured frames were scaled down to smaller sizes and the effect of resizing was tested for various image sizes. An image size of 512x512 is sufficient for a consistently accurate power line detection The program developed by Assistant Professor Cihan Topal from Anadolu University Electrical and Electronics Department was used to draw ground truths.

Files

Categories

Computer Vision, Image Processing, Machine Learning, Line Detection, Pattern Recognition, Pattern Detection

Licence