T-Rex Label
verifiedFreemiumtrexlabel.com
Browser-based zero-shot labeling tool for CV teams needing fast bounding-box and segmentation annotation without fine-tuning.
What it does
T-Rex Label is a browser-based annotation tool built on an open-set detection model. Users draw a box or prompt around one object and it automatically finds and labels similar objects across an image set without any fine-tuning.
Core features
Zero-shot open-set object detection, no fine-tuning needed
One-prompt batch annotation across multiple images
Smart segmentation and bounding-box tools
Cross-image reasoning to replicate a single annotation prompt
No install, fully browser-based
Import/export compatible with Kaggle, Roboflow, Hugging Face, Label Studio, and common ML frameworks
Best for
→Computer vision engineers building datasets for rare or novel object classes
→Agriculture, retail, or logistics teams labeling domain-specific imagery
→Teams wanting fast dataset creation without maintaining custom-trained annotators
Toolspool rankingby monthly traffic