Sigui Depin 1M: Multichain Transaction Graph Images for AI Security

Sigui Depin 1M contains 1,000,000 visual transaction graph images generated from 1.87 million real on-chain transactions across Ethereum, Arbitrum, and Polygon. Each image is annotated with an attack topology label designed to train Vision-Language Models to detect DeFi threats. The dataset was authored by Ibonon and last updated on HuggingFace in May 2026.

Use Cases

• Train Vision-Language Models to classify attack topologies based on annotated transaction graph images.
• Benchmark anomaly detection algorithms for DeFi security using labeled visual data.
• Research cross-chain transaction patterns and their association with known attack types.

Strengths

• Contains 1,000,000 annotated images, providing a large-scale resource for model training.
• Derived from 1.87 million real transactions across three major blockchain networks: Ethereum, Arbitrum, and Polygon.
• Annotations are specifically designed for training Vision-Language Models on DeFi threat detection.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Last updated 2026-05-09 05:35:38; freshness should be verified.

Provenance

Source

HuggingFace dataset authored by Ibonon.

Collection Method

Images generated from real on-chain transactions across Ethereum, Arbitrum, and Polygon.

Freshness

Last updated 2026-05-09 05:35:38.