Full lifecycle with an intuitive SME interface
Run Geti™ as a native Windows application or a pre-built Docker image for Linux. Annotate, train, optimize, deploy, and continuously improve models in one interface.
Geti™ is an open-source, end-to-end platform for building and continuously improving your own computer vision models. It covers annotation, fine-tuning, optimization, deployment, and inference — with 1,600+ model variants for detection, classification, and segmentation, optimized by OpenVINO™ for Intel® CPUs, GPUs, and NPUs.
Choose the Geti™ application for an intuitive no-code experience or its getitune engine for code-first workflows — both paths lead to production-ready OpenVINO™ models optimized for Intel® hardware.
Full lifecycle with a simple Python API
Use getitune to select a model, point it to your dataset, and train and optimize it on Intel® CPUs or GPUs with just a few lines of Python.
Start Building AI with Geti™
Pick your preferred installation method to get up and running locally.
1600+ Model Variants, Out of the Box
Geti™ gives you access to more than 1600 model variants across the three core computer vision tasks, built on a curated selection of 97 state-of-the-art architectures — from ultra-fast mobile networks to large transformer backbones. Every model is automatically exported to OpenVINO™ IR for efficient edge inference, so you can pick the right trade-off between speed, accuracy, and footprint without leaving the platform.
Object detection
The widest selection in Geti™: YOLO variants (YOLOX, YOLO11, YOLO12, YOLO26), real-time DETRs (D-FINE, DINOv3-DETR, RF-DETR, RT-DETR, ECDet), and mobile-friendly detectors for any deployment target.
- ATSS MobileNet V2
- SSD MobileNet V2
- YOLOX Tiny
- YOLOX S
- YOLOX L
- YOLOX X
- RT-DETR R50
- D-FINE M
- D-FINE L
- D-FINE X
- DINOv3 DETR S
- DINOv3 DETR M
- DINOv3 DETR L
- RF-DETR N
- RF-DETR S
- RF-DETR M
- RF-DETR L
- ECDet S
- ECDet M
- ECDet L
- ECDet X
- YOLO11 N
- YOLO11 S
- YOLO11 M
- YOLO11 L
- YOLO11 X
- YOLO12 N
- YOLO12 S
- YOLO12 M
- YOLO12 L
- YOLO12 X
- YOLO26 N
- YOLO26 S
- YOLO26 M
- YOLO26 L
- YOLO26 X
Instance segmentation
Pixel-perfect masks with classic Mask R-CNN backbones, the full RF-DETR size range, and YOLO11/YOLO26 segmentation variants for high-accuracy or real-time needs.
Image classification
From compact mobile CNNs to self-supervised transformer backbones and the YOLO26 classification family — pick the right balance of accuracy and latency. Any timm backbone can also be plugged in for fine-tuning.
- ViT Tiny
- ViT Small
- ViT Base
- ViT Large
- DINOv2 Small
- DINOv2 Base
- DINOv2 Large
- DINOv2 Giant
- EfficientNet B0
- EfficientNet B1
- EfficientNet B2
- EfficientNet B3
- EfficientNet B4
- EfficientNet B5
- EfficientNet B6
- EfficientNet B7
- EfficientNet V2 Small
- EfficientNet V2 Medium
- EfficientNet V2 Large
- MobileNet V3 Small
- MobileNet V3 Large
- ResNet18
- ResNet34
- ResNet50
- ResNet101
- ResNet152
- ConvNeXt Tiny
- ConvNeXt Small
- ConvNeXt Base
- ConvNeXt Large
- Swin Tiny
- Swin Small
- Swin Base
- DeiT Tiny
- DeiT Small
- DeiT Base
- YOLO26 N
- YOLO26 S
- YOLO26 M
- YOLO26 L
- YOLO26 X
- 1600+ backbones
Variant count includes the timm backbones available for classification fine-tuning.
End-to-End Workflow
Geti™ connects your model to the real world from both ends. Collect images, video, or live camera streams; accelerate annotation with smart assistants; then wire predictions back out through inference pipelines — all in one continuous workflow around your model.
Deploy Anywhere
Under the hood, Geti™ uses the OpenVINO™ toolkit to optimize model performance for deployment across the entire Intel® hardware portfolio. Pick the precision that fits your use case — INT8, FP16, or FP32 — and export to the runtime that matches your target, ready to run at the edge or in the cloud.
