Open Hardware Benchmark & Lineage Registry for Low-Energy Edge AI
“If I have seen further, it is by standing on the shoulders of giants.” — Isaac Newton
Open-source OBB detection + tracking for fixed perimeter, local NVR, and edge vision — scored on real low-energy ARM silicon.
Season Zero100% Open-Source Benchmark
- SPDX LicensesApache · MIT · BSD · ISC · CC-BY
- Author OwnershipOriginal License Preserved
- RecognitionPodium Badges & Lineage
Zero entry fees, zero commercial strings. Non-commercial and excluded models are screened out. Open-Source Policy →
Accounts start with 3 free Run Tokens and regenerate +1 every 24h (up to 3 stored). Tokens are spent only after Stage A manifest, license, and static ONNX checks pass.
Hardware & Early Partners: Season Zero is an early air-gapped HIL pilot — if you want to benchmark models on your own board/NPU, co-design a future workload track, or run a private hardware pilot, see how to work with us →
Submit an Open-Source Model
Submit a public GitHub or Hugging Face repository with your ONNX model and declarative manifest.
- Repository LayoutGitHub / HF Repo + model.json + .onnx
- Hardware Envelope≤40MB Artifact · ≤800MB RAM · ≥2.4 FPS
- Placement Floor≥25% OBB mAP@50:95 · ≥30% Track HOTA · ≥35 Domain
Point Claude Code, Codex, Cursor, or Antigravity at /llms.txt. Your agent builds & verifies the ONNX model locally with verify_local.py, pushes to your repo, and hands you a pre-filled review link so you approve and spend 1 Run Token with a single click.