plitip/team-qh-091-safesite
object-detection model by plitip
From the publisher
Model card & documentation
Source previewRead the publisher’s intended use, setup instructions, evaluations and limitations. The original model card is the source of truth.
Fine-tuned Ultralytics YOLO26 detectors for construction-site photos. They are the trained detector inside Salamtak سلامتك ("your safety"), built at the Qai × Hugging Face Hackathon (Doha, 5–7 Nov 2026). A plain-Python rules engine (no LLM) turns detections into hazards, a 0–100 safety score and multilingual worker alerts. The Space runs v2n (v2n/best.pt, YOLO26n, pinned to commit b3fac75b643c0508f54ecccb045a385ce8798221). This is a deliberate captain override of Gate 3. The team's rule (Gate 3) lets v2 replace v1 only if helmet and nohelmet on the original v1 val images stay within 0.02 of v1 and person, vest and novest improve on fully labelled images (plus two app-level checks: hazard catch rate on our Qatar test photos and CPU time per photo). For the deployed v2n: The captain overrode the failed nohelmet check: v2n trades a small drop on nohelmet (−0.031 vs. v1) for much better whole-person, vest, machinery and vehicle detection, the new…
Read the full model card ↗ · Preview checked 2026-11-08T07:05:28.812Z
Inside the original model card — Document outline
- Salamtak سلامتك — PPE & site-hazard detector (YOLO26)
- Which weights the Space uses
- v2 — 11 classes (YOLO26s, trained 7 Nov 2026)
- Classes
- Results
- v2n — YOLO26n (smaller and faster; the deployed model)
- Training data (v2)
- Caveats (v2)
- v1 — 7 classes (YOLO26s, trained 6 Nov 2026)
- Classes
- Results
- Training data and licenses (v1)
Headings are captured from the source. Links open the publisher’s document, not a locally hosted copy.
Links from the model card
References supplied by the publisher, not independently verified endorsements. Check the destination before downloading files or entering credentials.
Documentation belongs to its respective authors. Reported project/model license: agpl-3.0. A listing is not a grant of reuse or training rights. Confirm the document’s own terms at the source.
Model overview
plitip/team-qh-091-safesite is a object-detection model repository published by plitip on Hugging Face. The source reports the ultralytics library.
This page summarizes Hub metadata. For intended use, training data, evaluation results and limitations, consult the original model card.
Model facts
- Task
- object-detection
- Library
- ultralytics
- Recent downloads (30 days)
- 1,555
- Cumulative likes
- 0
- Hugging Face trending score
- 0
- Reported safetensors parameters
- Not reported
- Architecture
- Not reported
- License
- agpl-3.0
- Access
- Not gated by Hugging Face
- Created
- 2026-11-06T15:44:12.000Z
- Last modified
- 2026-11-07T07:41:31.000Z
Compatibility and lineage
Language coverage was not reported.
Base-model lineage was not included.
Parameter count is not a RAM/VRAM requirement. Check precision, quantization, context length and runtime compatibility in the model card; no hardware or API-cost claim is inferred here.
Use and evaluate
Check license terms and access requirements first. Review the model card’s documented loading instructions, evaluations and safety limitations. Benchmark scores are not imported by this directory, and popularity is not an accuracy ranking.
No model weights are downloaded or executed by AltAPIs. Never enable remote model code without reviewing it.
Source and freshness
Source: Hugging Face Hub. Metadata observed 2026-11-08T06:59:29.904Z. Daily imports are snapshots, not real-time monitoring.
Popularity and source listings do not establish security, suitability, licensing rights or benchmark performance.
Related models
PekingU/rtdetr_r101vd_coco_o365
microsoft/table-transformer-structure-recognition
PaddlePaddle/PP-DocLayoutV3_safetensors