safesight-edge/include
sladro aaa794fe10 feat: Enhance preprocessing and detection with letterbox resizing and transformation metadata
- Added `resize_mode` option to `preprocess_node` for handling image resizing strategies: stretch, keep ratio, and letterbox.
- Implemented `BuildLetterbox` function to create letterbox resized frames while maintaining aspect ratio.
- Introduced `FrameTransformMeta` structure to store transformation metadata for frames, including scaling and padding information.
- Updated detection coordinate calculations in `ai_yolo_node` to utilize transformation metadata for accurate bounding box adjustments.
- Enhanced bounding box mapping in `osd_node` to account for potential discrepancies between detection and frame sizes.
- Refactored image processing logic to support new resizing modes and ensure compatibility with existing formats.
2026-02-27 20:05:20 +08:00
..
face 测试人脸识别功能,刚刚开发完成 2026-01-07 17:15:07 +08:00
frame feat: Enhance preprocessing and detection with letterbox resizing and transformation metadata 2026-02-27 20:05:20 +08:00
hw 解耦开发6 2026-01-16 20:29:32 +08:00
media 优化性能 2026-01-06 16:26:39 +08:00
utils 增加稳定性 2026-01-29 20:44:16 +08:00
ai_scheduler.h Refactor YOLO node output processing and enhance output management in AI scheduler 2026-02-27 19:39:08 +08:00
graph_manager.h 解耦开发3完成,准备编译 2026-01-16 19:10:30 +08:00
http_server.h 开发sprint6阶段,第一次开发 2025-12-31 09:40:36 +08:00
media_server_app.h 性能优化4 2026-01-12 20:51:27 +08:00
node.h 解耦开发3完成,准备编译 2026-01-16 19:10:30 +08:00
plugin_loader.h 规范优化4 2026-01-13 08:28:55 +08:00
version.h Initial commit: OrangePi3588 Media Server project 2025-12-08 17:05:22 +08:00