AED-Net: A High-Resolution Remote Sensing Image Road Extraction Method Integrating Atrous Spatial Pyramid Pooling and Efficient Channel Attention Mechanism
Bibliographic record
Abstract
High-resolution remote sensing imagery plays a pivotal role in road extraction due to its abundant semantic information. However, traditional road extraction methods are time-consuming, labor-intensive, and susceptible to subjective factors, leading to inaccurate extraction results. Therefore, this study aims to explore an efficient, accurate, and automated road extraction method. We propose a road extraction model named AED-Net, which employs a lightweight MobileNet v2 as the feature extractor, combines the multi-scale feature extraction capability of ASPP with an encoder-decoder structure, and integrates an ECA mechanism to enhance feature learning ability. During the training process, the BCE-DL loss function is adopted to optimize model performance. Experimental validation on the Massachusetts Roads dataset and Ottawa Road dataset demonstrates that AED-Net outperforms the latest Attention U-Net (AU-Net) and ECA-DeeplabV3+ models. On the Massachusetts Roads dataset, AED-Net achieves an overall accuracy (OA) of 97.60% and a mean intersection over union (MIoU) of 85.29%. Similarly, on the Ottawa Road dataset, AED-Net exhibits superior performance compared to classical semantic segmentation networks such as SegNet, U-Net, and Deeplab V3+, achieving an OA of 98.83% and an MIoU of 88.74%. Furthermore, ablation studies validate the effectiveness and necessity of the proposed improvements. In summary, AED-Net has made significant progress in multi-scale feature extraction, boundary information restoration, and critical feature learning, offering new insights and solutions for the development of road network extraction technologies. Its high accuracy and robustness render AED-Net promising for a wide range of applications in remote sensing image processing, intelligent transportation systems, and beyond.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.001 |
| Meta-epidemiology (narrow) | 0.002 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.002 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.002 |
| Open science | 0.002 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.004 | 0.002 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".