REVIEW 1 cited by
DUFormer: Solving Power Line Detection Task in Aerial Images using Semantic Segmentation
Not yet reviewed by Pith; the record is open.
This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.
SPECIMEN: schema-true, not a live event
T0 review · schema-true
One-sentence machine reading of the paper's core claim.
pith:XXXXXXXX · record.json · timestamp
DUFormer: Solving Power Line Detection Task in Aerial Images using Semantic Segmentation
read the original abstract
Unmanned aerial vehicles (UAVs) are frequently used for inspecting power lines and capturing high-resolution aerial images. However, detecting power lines in aerial images is difficult,as the foreground data(i.e, power lines) is small and the background information is abundant.To tackle this problem, we introduce DUFormer, a semantic segmentation algorithm explicitly designed to detect power lines in aerial images. We presuppose that it is advantageous to train an efficient Transformer model with sufficient feature extraction using a convolutional neural network(CNN) with a strong inductive bias.With this goal in mind, we introduce a heavy token encoder that performs overlapping feature remodeling and tokenization. The encoder comprises a pyramid CNN feature extraction module and a power line feature enhancement module.After successful local feature extraction for power lines, feature fusion is conducted.Then,the Transformer block is used for global modeling. The final segmentation result is achieved by amalgamating local and global features in the decode head.Moreover, we demonstrate the importance of the joint multi-weight loss function in power line segmentation. Our experimental results show that our proposed method outperforms all state-of-the-art methods in power line segmentation on the publicly accessible TTPLA dataset.
Forward citations
Cited by 1 Pith paper
-
SEMIR: Topology-Preserving Graph Minors for Thin-Structure Segmentation
SEMIR uses topology-preserving graph minors to collapse pixel lattices into small supernode graphs for GNN classification, enabling full-resolution thin-structure segmentation without patching and matching baselines o...
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.