Deep learning applied to urban agriculture: spatial-temporal changes of agricultural land in a rapidly urbanizing Southeast Asian city
Bibliographic record
Abstract
Rapid urbanization in Southeast Asia has been posing huge impacts on local food systems, altering spatial-temporal patterns of urban agriculture, ecosystems and social life. Understanding these changes is crucial for cities planning their land use and infrastructure development to achieve a balance between urban growth, agricultural sustainability, and food security. This study mapped the changes between 2013 and 2020 of five agricultural types within (peri)urban areas in Huế, a province’s capital in Vietnam. High-resolution SPOT satellite images (1.5m) and a deep learning model based on the U-net architecture were used to map land use and agriculture types. This approach addresses challenges in generating extensive labelled datasets in urban settings characterized by fragmented farmland and dense development. The optimized U-net model achieved high classification performance (for 2013: IoU = 0.86 and Kappa = 0.93, for 2020: IoU = 0.87 and Kappa = 0.92) even when operated on regular CPU computers, demonstrating its practical applicability for countries with limited technical infrastructure. This is also the first study in Southeast Asia to accurately map (overall accuracy 85% for 2013 and 87% for 2020) multiple types of urban agriculture at 1.5 m resolution, enabling detailed spatial-temporal changes analysis. These results can inform decision-makers in elaborating effective land use strategies and food security plans, and offer researchers a scalable deep learning framework for urban agriculture mapping in rapidly urbanizing regions.
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) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.000 | 0.001 |
| Research integrity | 0.000 | 0.001 |
| Insufficient payload (model declined to judge) | 0.001 | 0.000 |
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".