SendTheCatCATVISHThe All-in-One Desktop
Computer Vision Workspace
A local-first, Tauri-powered IDE. Manage datasets, annotate with AI, train YOLO models, and build inference flows visually—all from one lightning-fast app.
Built for Speed and Structure
Forget messy Python scripts and scattered folders. Catvish organizes your computer vision projects into a unified, version-controlled workspace.
Dataset Management
Recursively import images, detect duplicates, and organize classes. View your data distribution instantly.
Versioning
Create immutable dataset versions. Track exactly which version was used to train "Model v2.0". No more "final_final_v3".
Smart Annotation
Built-in labeler with keybinds for speed. (Coming Soon: AI-assisted labeling).
Train. Optimize. Deploy.
A fully integrated MLOps pipeline running locally on your machine.
One-Click Training
Launch YOLOv8 training runs directly from the UI. Visualize loss curves, mAP, and metrics in real-time without Context Switching.
Export & Compress
Convert your best weights to ONNX or OpenVINO. Apply FP16/INT8 quantization for lightning-fast inference performance.
Instant Serving
Spin up a local API endpoint or test in the visual Playground. No Docker, no Kubernetes—just works.
-F "image=@cat.jpg"
{
"class": "cat",
"conf": 0.98
}
Visual Inference Flow
Chain models and logic together in a node-based editor. Create complex pipelines like "Detect Person -> Check PPE -> Log Violation" without writing code.