更新日志

此页面记录 YOLO-Toys 各版本的变更。

All notable changes to this project will be documented in this file.

The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.

---

[Unreleased]

• Placeholder for upcoming features

---

3.1.0 (2026-04-16)

📝 Documentation System Refactor

This release focuses on comprehensive documentation improvements and bilingual support.

• Professional bilingual documentation system (English/Chinese)

• Restructured docs/ directory with organized categories:

  • getting-started/ — Installation and quick start guides
  • api/ — REST API and WebSocket protocol documentation
  • architecture/ — System design and handler pattern documentation
  • deployment/ — Docker and production deployment guides
  • guides/ — Development tutorials (adding custom models)
  • reference/ — Quick reference materials (models, FAQ)

• Standardized changelog following Keep a Changelog format

• Complete API reference with request/response examples

• Architecture documentation explaining design patterns (Strategy, Registry, Template Method)

• Model compatibility matrix with performance benchmarks

• Comprehensive FAQ covering installation, performance, and troubleshooting

• Documentation navigation optimized with clear categorization

• Cross-language linking between English and Chinese documents

• Enhanced code examples in all documentation

Documentation

• All documentation now available in both English and Chinese
• README.md and README.zh-CN.md cross-referenced
• Changelog follows Keep a Changelog specification

---

3.0.0 (2026-02-13)

🏗️ Architecture Refactor

Major architectural overhaul with strategy pattern implementation and improved maintainability.

• Strategy Pattern Architecture: New app/handlers/ module

  • base.py — BaseHandler abstract class with unified interface
  • registry.py — HandlerRegistry factory and MODEL_REGISTRY metadata
  • yolo_handler.py — YOLO detection/segmentation/pose handler
  • hf_handler.py — DETR / OWL-ViT / Grounding DINO handlers
  • blip_handler.py — BLIP Caption / VQA handlers

• Pydantic Settings: Unified environment configuration in app/config.py

  • Type validation and .env file support
  • Derived properties for computed settings

• Enhanced Testing: Test cases increased from 6 to 16+

  • WebSocket connection/inference/config update tests
  • Model info endpoint tests (404/200)
  • Boundary input tests (invalid content-type, empty files)
  • HandlerRegistry unit tests
  • AppSettings configuration tests

• Modern FastAPI Patterns

  • Lifespan context manager replacing deprecated @app.on_event
  • Route extraction to dedicated routes.py
  • Structured logging replacing print statements

• model_manager.py simplified from 814 lines to ~100 lines

• main.py reduced from 410 lines to 91 lines

• Frontend URL building extracted to common functions

• .gitignore and .dockerignore enhanced

• pyproject.toml updated with [project] metadata

Removed

• Deprecated on_event("startup") usage
• Redundant code in model manager

---

2.0.0 (2025-11-25)

🎯 Multi-Model System

Comprehensive support for multiple AI models with dynamic switching.

• Multi-Model Support System

  • YOLO Detection: YOLOv8 n/s/m/l/x
  • YOLO Segmentation: YOLOv8-seg series
  • YOLO Pose: YOLOv8-pose series
  • Transformer: DETR ResNet-50/101
  • Open Vocabulary: OWL-ViT zero-shot detection
  • Multimodal: BLIP Caption and Visual QA

• New API Endpoints

  • GET /models — List all available models by category
  • GET /models/{model_id} — Get model details
  • POST /caption — Image captioning
  • POST /vqa — Visual question answering
  • WebSocket configuration updates at runtime

• Frontend UI Overhaul

  • Modern dark/light theme design
  • Model category tabs with quick switching
  • Toast notification system
  • Real-time performance overlay
  • Settings persistence in localStorage

• Model management unified through ModelManager

• API response format standardized

• Frontend refactored with ES Modules

---

1.0.0 (2025-02-13)

🚀 Project Launch

Initial project infrastructure and basic YOLO functionality.

• FastAPI backend with REST endpoints
• YOLOv8 object detection support
• WebSocket streaming for real-time detection
• Basic HTML/CSS/JS frontend
• Docker support
• GitHub Actions CI/CD

---

Release Link Format