Satellite Imagery for Brick Kiln Detection in South Asia

SentinelKilnDB is a large-scale benchmark dataset for oriented bounding box detection of brick kilns using satellite imagery. It was created by SustainabilityLabIITGN for the NeurIPS 2025 Datasets & Benchmarks Track, focusing on the Indo-Gangetic Plain region of South Asia.

Use Cases

• Train object detection models on satellite imagery to identify brick kiln locations using oriented bounding boxes.
• Benchmark detection algorithms for geospatial targets like brick kilns across the South Asia region.
• Analyze the spatial distribution of brick kilns in the Indo-Gangetic Plain from satellite data.

Strengths

• Dataset is designated for the NeurIPS 2025 Datasets & Benchmarks Track, indicating a research standard.
• Focuses on a specific, impactful application: detecting pollution sources in a highly polluted region.
• Described as a large-scale dataset for a specialized computer vision task in geospatial analysis.

Limitations

• Specific scale metrics like image count, annotation count, or resolution are not provided in the input.
• Geographic coverage is limited to South Asia, specifically the Indo-Gangetic Plain.
• Dataset details such as columns, sample data, and file formats are unknown from the provided input.

Provenance

Source

SustainabilityLabIITGN via Hugging Face.

Collection Method

null

Time Range

null

Freshness

Last updated on October 7, 2025.

Geography

South Asia, specifically the Indo-Gangetic Plain spanning northern India, Pakistan, and Bangladesh.