Automatic High Resolution Wire Segmentation and Removal

Chiu, Mang Tik; Zhang, Xuaner; Wei, Zijun; Zhou, Yuqian; Shechtman, Eli; Barnes, Connelly; Lin, Zhe; Kainz, Florian; Amirghodsi, Sohrab; Shi, Humphrey

doi:10.1109/cvpr52729.2023.00217

Cited by 2 publications

(3 citation statements)

References 30 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This preliminary investigation paves the way for future research aimed at expanding the scope of our approach to diverse tasks and applications beyond traditional road segmentation. We employed the wire dataset described in [11]. A dataset of 420 copyright-free test images along with annotations for public use and evaluation is available, we split the datasets in 2-fold (50% training and 50% test sets).…”

Section: E Different Applicationmentioning

confidence: 99%

“…Moreover, the combination of these functions obtains good performance and provides an interesting yet initial starting point for future work. We are unable to directly compare with [11] because they utilize a significantly larger dataset that is not accessible.…”

Section: E Different Applicationmentioning

confidence: 99%

“…Comparison of the loss functions in terms of the different metrics (in %). Performance of a single network on the wire dataset[11].…”

mentioning

confidence: 99%

See 2 more Smart Citations

An Enhanced Loss Function for Semantic Road Segmentation in Remote Sensing Images

Nanni,

Brahnam,

Loreggia

2024

IEEE Access

View full text Add to dashboard Cite

The analysis of road continuity in satellite images is a complex challenge. This is due to the difficulty in identifying the directional vector of road sections, especially when the satellite view of roads is obstructed by trees or other structures. Today, most research focuses on optimizing the deep learning network topology, however, the accuracy of segmentation is affected by the loss function used in training; currently, little research has been published on ad-hoc loss functions for road segmentation.To solve this problem, we proposed loss functions based on topological pixel analysis, in which more weight is given to problematic pixels representing non-real road breaks. We report the results of different tests, obtaining state-of-the-art performance among convolution neural network based approaches. Both the code and information for replicating our experiments are available on https://github.com/LorisNanni/An-Enhanced-Loss-Function-for-Semantic-Road-Segmentation-in-Remote-Sensing-Images , so as to enable future reliable comparisons.

show abstract

Section: E Different Applicationmentioning

confidence: 99%

Section: E Different Applicationmentioning

confidence: 99%