visionfamily
RF-DETR
Roboflow real-time detection transformer family (N/S/M/L/XL/2XL); ICLR 2026 poster. Segmentation support per Roboflow is unverified.
Levelfamily
Classvision
Parametersnot in the notes
Contextnot in the notes
ReleasedNov 2025
Opennessunknown
Runs onsingle datacentre GPU, embedded GPU
Statusactive
Benchmark Results
vendor-reported
COCO val valdomain
Where this comes from
Every score names who ran it: vendor-reported, maintainer-run or independent. A rank is shown with the population it was ranked in (gap rule 3).
Lineage
successorfinetune
LW-DETR is the parent (successor)medium confidencemodel card
DINOv2 is the parent (finetune)medium confidencemodel cardRF-DETR 2XL belongs to this familyvisionNov 2025126.9MRF-DETR L belongs to this familyvisionNov 2025RF-DETR M belongs to this familyvisionNov 202533.7MRF-DETR N belongs to this familyvisionNov 202530.5MRF-DETR S belongs to this familyvisionNov 202532.1MRF-DETR XL belongs to this familyvisionNov 2025Where this comes from
Parent and child relations from model cards, Hugging Face base_model fields and the genealogy tables of the research notes. The tree shows two generations each way, 5 models. Confidence and evidence are the notes' own.
Identity
Where this comes from
Developer and dates from the research notes. Release date is the vendor's public announcement; weights date is when open weights were published.
Architecture and Size
- Architecture
- dense transformer Real-time DETR; LW-DETR base with DINOv2 backbone replacing CAEv2; neural architecture search
- Total parameters
- – sources
- Active parameters
- –
- Checkpoint parameters
- –
- Note
- N 30.5M, S 32.1M, M 33.7M, 2XL 126.9M
- Context length
- not in the notes
- Input
- image
- Output
- image, text
Where this comes from
Parameter counts as the developer states them. Where the checkpoint holds a different total, both are shown (register A16).
Openness and Access
- Openness
- unknown
- Licence
- not in the notes sources
- Access routes
- open weights download
Where this comes from
The licence file governs over registry tags and aggregator labels.
Deployment
- Runs on
- single datacentre GPU, embedded GPU
- Footprint
- N 2.3 ms, 2XL 17.2 ms latency
- Runtimes
- not in the notes
Where this comes from
Deployment tiers and runtimes as named in the notes. Footprint is the notes' figure, not a measurement by AURION.
Indonesia
- Indonesian support
- not stated
- Regional languages
- not in the notes
- Jakarta hosting
- unknown
- Languages
- not in the notes
Where this comes from
Indonesia fields are carried on every model; unknown is an allowed value.
Industry Use Cases
No industry use case links this model.
Links
DeveloperRoboflowcompany, USA
Roboflowcompany, USAModelRF-DETRRoboflow, vision
RF-DETRRoboflow, visionUse CasesSafety (PPE, Hazards, Fatigue)IndustrialVisual Quality Inspection and Defect DetectionIndustrialUse cases of models in the same R&D topic; none is recorded for this model itself.
Who Runs It
PT Bukit Asam TbkIn production, 9 Sep 2026
PT Aneka Tambang Tbk (ANTAM), ferronickelIn production, 9 Sep 2026PLN Indonesia Power with Huawei (PLTU Banten 3 Lontar)Research, 26 Nov 2025
PT Kilang Pertamina Internasional (Pertamina refineries)In production, 17 Oct 2025+5 moreDeployments of these use cases; none names this model.
PT Bukit Asam TbkIn production, 9 Sep 2026PT Aneka Tambang Tbk (ANTAM), ferronickelIn production, 9 Sep 2026PLN Indonesia Power with Huawei (PLTU Banten 3 Lontar)Research, 26 Nov 2025PT Kilang Pertamina Internasional (Pertamina refineries)In production, 17 Oct 2025+5 moreDeployments of these use cases; none names this model.Models 8
Organisations 1
Sources
- RF-DETR: Neural Architecture Search for Real-Time Detection Transformers arxiv.org https://arxiv.org/html/2511.09554v2
- ICLR Poster RF-DETR: Neural Architecture Search for Real-Time Detection Transformers iclr.cc https://iclr.cc/virtual/2026/poster/10007257
To cite this page: Project AURION, AI Use-Cases Research & Innovation, "RF-DETR", https://aurion.semai.id/models/roboflow-rf-detr/, records of October 2026.