rf-detr-medium
by Roboflow
Balanced real-time detection transformer for object metadata pipelines
Roboflow/rf-detr-mediummixpeek://image_extractor@v1/roboflow_rf_detr_medium_v1Overview
RF-DETR Medium is the balanced checkpoint in Roboflow's real-time detection transformer family. It keeps the DETR-style end-to-end detection formulation while reducing the footprint compared with the Large model.
On Mixpeek, RF-DETR Medium is a practical default for high-volume video and image pipelines where agents need object metadata for filtering, counting, and follow-up visual inspection without paying for the largest detector on every frame.
Architecture
Detection transformer with a ViT-style backbone, multi-scale projection, deformable cross-attention decoder, and DETR-style object queries. The medium checkpoint targets the throughput-quality middle ground.
Mixpeek SDK Integration
import { Mixpeek } from "mixpeek";const mx = new Mixpeek({ apiKey: "API_KEY" });await mx.collections.ingest({collection_id: "warehouse-video",source: { url: "s3://warehouse/camera-01.mp4" },feature_extractors: [{feature: "object_detection",model: "Roboflow/rf-detr-medium"}]});
Capabilities
- COCO-style object detection
- Transformer-based boxes without anchors
- Balanced latency and quality for production frame pipelines
- Apache 2.0 license
Use Cases on Mixpeek
Common Pipeline Companions
Explore on Mixpeek
Compare alternatives in this category
Hand-picked tools & platforms compared
Deep-dive technical guide
See how Mixpeek runs models as extractors
Store & search embeddings at scale
Usage-based pricing for pipelines
Compare models, APIs & infrastructure
Specification
Research Paper
RF-DETR Medium
arxiv.orgBuild a pipeline with rf-detr-medium
Add this model to a processing pipeline alongside other extractors. Combine with retrieval stages for end-to-end search.
Open Studio