3 Commits

Author SHA1 Message Date
4f9a193782 chore(local-tool): 移除模型 FPS/latency/accuracy 預估值
使用者要求拿掉 FPS 預估數字(未經實測,不準確)。

- docs/LOCAL-TOOL-SPEC.md: 模型表格移除 FPS 欄位
- server/data/models.json: 7 個模型全部移除 fps / latencyMs / accuracy 欄位

前端 model-detail / model-card 有讀這些欄位的 UI,移除後會顯示 — 或不顯示
該列,不需要額外前端改動(已有 null/undefined fallback)。

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-04-16 15:01:25 +08:00
9793a2efc1 chore(local-tool): models.json 只保留有實體 .nef 的 7 個預設模型
使用者要求:前端模型庫只保留有適配裝置的模型(KL520 4 個 + KL720 3 個)。

原本 models.json 有 15 筆:
  - 8 筆 ONNX framework 的 demo 模型(yolov5-face-detection /
    imagenet-classification / person-detection / vehicle-classification /
    hand-gesture-recognition / coco-object-detection / face-mask-detection /
    license-plate-detection)— 都沒有實體 .nef 檔,是 placeholder metadata
  - 7 筆 NEF framework 的實體模型(每個都有 filePath 指向 data/nef/)

現在只保留 7 筆實體模型:
  KL520(4 個):
    - kl520-yolov5-detection (yolov5 no-upsample 640x640)
    - kl520-fcos-detection (fcos-drk53s 512x512)
    - kl520-ssd-face-detection (ssd_fd_lm 320x240)
    - kl520-tiny-yolov3 (tiny_yolo_v3 416x416)
  KL720(3 個):
    - kl720-yolov5-detection (yolov5 no-upsample 640x640)
    - kl720-resnet18-classification (resnet18 224x224)
    - kl720-fcos-detection (fcos-drk53s 512x512)

server/data/models.json 是 runtime 讀取,三平台(macOS/Windows/Linux)
共用同一份,改一次三平台全部生效。

驗證:
- python3 json.load 解析正常,7 筆 entries
- 每筆 filePath 指向的 .nef 實體檔都存在於 server/data/nef/{kl520,kl720}/
- 檔案大小:1-13 MB,合計 ~64 MB
- macOS dmg 重 build 163MB OK
- Bundle 內 Contents/Resources/data/models.json 更新為 7 筆

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-04-16 08:44:16 +08:00
c54f16fca0 Initial commit: visionA monorepo with local-tool subproject
local-tool/: visionA-local desktop app
- M1: Wails shell + Go server + Next.js UI + Mock mode (macOS dmg ready)
- M2: i18n (zh-TW/en) + Settings 4-tab refactor
- M3: Embedded Python 3.12 runtime (python-build-standalone) + KneronPLUS wheels
- M4: Windows Inno Setup script (build on Windows runner)
- M5: Linux AppImage script + udev rule (build on Linux runner)
- M6: ffmpeg (GPL, pending legal review) + yt-dlp bundled
- Lifecycle: watchServer health check, fatal native dialog,
            Wails IPC raise endpoint, stale process cleanup

.autoflow/: full PRD / Design Spec / Architecture / Testing docs
            (4 rounds tri-party discussion + cross review)
.github/workflows/: macOS / Windows / Linux build CI

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-04-11 22:10:38 +08:00