JIET Studio is a high-performance, modular environment designed for professional-grade YOLO training, dataset management, and real-time inference. Engineered for speed and precision, it streamlines the entire computer vision workflow from raw data to production-ready models.
Experience a seamless end-to-end pipeline with our specialized toolset:
The Precision Labeling Engine. A high-velocity annotation tool supporting Edit, Draw, and Magic Wand (SAM-powered) modes. Designed for rapid, accurate data labeling with zero-friction workflows.
Professional Training Orchestrator. A robust interface for training YOLO models. Features real-time progress parsing, advanced hyperparameter tuning (Mosaic, Mixup, Blur), and one-click resumption.
Modular Data Synthesis. A dynamic augmentation pipeline that allows for complex, multi-stage image transformations using a modular filter-loading system.
How To Make Custom Filters? Custom Filter Documentation
Real-Time Vision Analytics. A versatile inference engine for real-time model validation. Supports single images, batch folders, video files, and live webcam streams.
Dataset Utility Fabric. Essential tools for dataset hygiene, including specialized Health Checks, class distribution analysis, and intelligent video frame extraction.
Model Validation Dashboard. A comprehensive evaluation suite providing deep insights into model performance with side-by-side Ground Truth vs. Prediction visualizations.
- Setup Environment:
pip install -r requirements.txt - Launch:
python main.py - Configure: Visit the global Settings (JIET > Settings) to configure hardware acceleration and default models.
AGPL-3.0