# T-Rex Label Launches Browser-Based AI Annotation Tool for Computer Vision Interactive visual prompting and automated masking accelerate zero-shot dataset preparation directly inside browsers. - **Source:** [Dynamic Business](https://dynamicbusiness.com) - **Published:** 2026-10-14 - **Topic:** [Data & AI](https://brifing.si/c/data-ai) - **Original article:** - **On Brifing:** https://brifing.si/s/2febf73e-bf25-427c-a172-b212ad2335a8 ## Browser-Based Vision Annotation T-Rex Label has introduced a web-based artificial intelligence annotation tool designed to streamline training dataset development for computer vision applications. Operating fully within modern web browsers, the platform eliminates local software installation requirements while providing teams with specialized interactive tools to tackle tedious manual labeling workloads across dense visual scenes. ## Visual Prompt Object Detection The system features open-set detection driven by visual prompts, allowing annotators to draw a single bounding box over an exemplar target. The underlying model detects all similar object instances across the frame, supporting zero-shot recognition for rare categories or custom industrial items unseen during initial baseline model training. ## Automated Pixel-Level Segmentation Beyond standard rectangular bounding boxes, T-Rex Label incorporates automated mask annotation capabilities for generating precise pixel-level segmentations. Users can construct complex segmentation boundaries manually or leverage the integrated automatic mask engine, which significantly accelerates object boundary extraction during intensive image labeling workflows without demanding additional software installations. ## Text-Driven Automated Pre-Annotation For teams requiring larger-scale automation, the platform provides asynchronous pre-annotation triggered by text prompts. This background pipeline leverages large vision foundation models including DINO-X and Grounding DINO to classify and segment specified target categories across entire image repositories without requiring constant manual guidance from human operators. ## Batch and Cross-Image Propagation When users define a target object via visual prompting, T-Rex Label can propagate that label across all matching instances within the active image and throughout the broader dataset. This cross-image propagation enables fast batch processing, drastically cutting down repetitive labeling tasks across continuous image sequences and video frames. ## Broad Export Format Compatibility To ensure compatibility with existing machine learning training workflows, T-Rex Label exports completed annotations into standard industry structures. Supported export outputs include COCO, YOLO, TensorFlow, Hugging Face, ModelScope, Kaggle, and Label Studio formats, ensuring straightforward ingestion into diverse downstream model training environments and internal data pipelines. ## Tiered Usage and Pricing Model The platform provides community users with free, unlimited access to interactive annotation tools, including visual-prompt bounding boxes and automated masks. Conversely, asynchronous AI pre-annotation operates under a paid, credit-based model charged per processed image, though exact tier pricing tables require direct quote inquiries with the vendor. ## Deployment Impact for Vision Teams T-Rex Label targets machine learning teams and small enterprises building custom computer vision pipelines containing rare objects like agricultural pests or industrial components. By combining free interactive zero-shot tooling with standard format exports, the platform provides a practical path to iterate vision datasets without dedicated labeling infrastructure. ## Sources covering this story - https://dynamicbusiness.com/ai-tools/t-rex-label-browser-based-ai-annotation-for-accelerated-dataset-creation.html — --- _Summary written by Brifing from the original reporting. Copyright in the original article belongs to its publisher; read the full piece at the link above._