CFC26: Fish Detection and Tracking in Underwater Sonar Video

CFC26 is a large-scale benchmark dataset for fish detection, tracking, and counting in underwater ARIS sonar video. It is designed to evaluate generalization under distribution shift and deployment-relevant performance in ecologically diverse and acoustically challenging environments. The dataset was created by perona-lab and spans multiple river systems with substantial variation in fish size, density, sonar range, and background structure.

Use Cases

• Benchmarking fish detection models based on sonar video sequences.
• Evaluating model generalization under distribution shift across different river systems.
• Developing and testing fish tracking algorithms in acoustically challenging underwater environments.
• Training counting models for ecological population estimates based on sonar imagery.

Strengths

• Designed as a large-scale benchmark for evaluating generalization under distribution shift.
• Spans multiple river systems, capturing variation in fish size, density, and sonar range.
• Focuses on deployment-relevant performance in ecologically diverse environments.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Description metadata is limited; actual data quality requires manual inspection after download.

Provenance

Source

perona-lab via Hugging Face.

Collection Method

Likely collected from underwater ARIS sonar video across multiple river systems.

Freshness

Last updated 2026-05-06 07:03:01; freshness should be verified.

Geography

Multiple river systems.