Skip to content

VLM-AutoYOLO v1.5.7

Choose a tag to compare

View all tags
@github-actions github-actions released this 07 Jun 19:09
· 54 commits to master since this release
v1.5.7
f70b25c

What's Changed

  • style: ruff format dataset_import files (f70b25c)
  • feat: embed GPU passthrough into docker-compose.yml (90d3114)
  • docs: update STRUCTURE.md with new modules and services (c3a3ef9)
  • fix: replace PUT/DELETE with POST for chunk upload routes (d5571bd)
  • fix: add fileUrlCache WeakMap to deduplicate blob URLs, expose isRedetecting (fe18da8)
  • fix: destructure signal in useDetectMutation mutationFn params (998c40a)
  • docs: add v1.5.7 changelog, update READMEs with import/queue features (0f1e32c)
  • test: add 21 unit tests for dataset import parsers (0aeedd8)
  • style: ruff format import_dataset.py (fb2ff2d)
  • fix: initialize names dict before conditional in _read_yolo_names (d879c99)
  • fix: prevent dual active border on batch thumbnails (4f301dc)
  • fix: blob URL lifecycle - remove auto-revoke in Zustand, use useState+useEffect (6b6f16f)
  • fix: skip optimisticModelLoading when model is already loaded (a0cbf50)
  • fix: add missing setPreviewUrl destructuring in Home.tsx (4b99a3b)
  • fix: pending thumbnail selection uses cached blob URL, fix maskClosable deprecation (7b6817a)
  • fix: allow clicking pending batch thumbnails to preview image (844ab25)
  • fix: stop showing spinner on cancelled/ended batch detection thumbnails (af9bc28)
  • fix: stable blob URLs for batch thumbnails with ref-based caching (5361370)
  • fix: clear old result and batchResults on new detection, show first new image (47590d9)
  • fix: batch detection shows all images immediately with local blob previews (93b571b)
  • fix: chunk_init async body read, missing format in worker, i18n bbox/mask keys (e0b3336)
  • feat: chunked upload with Web Worker + resume support for dataset import (d1407dc)
  • chore: rewrite ja.json with kanji-heavy Japanese, minimize katakana (bf07bc0)
  • chore: sync ja.json i18n keys with zh/en - add cancel, import, delete confirm (2b5d034)
  • fix: remove duplicate closing brace in en.json (19bd092)
  • feat: dataset import from YOLO/COCO/VOC/CreateML ZIP archives (06eb321)
  • fix: allow queuing training jobs while another is running (3cd8cb8)
  • fix: add explicit Popconfirm import to avoid auto-import issue (be11422)
  • fix: replace window.confirm() with antd Popconfirm for delete/cancel actions (b018c8b)
  • feat: cancel in-flight detection requests with AbortController (acdf518)
  • feat: training job queue with cancel support (d124ac9)
  • fix: add confirm dialogs for destructive delete operations (23bdd61)
  • fix: SAM3 health check race condition and test skip logic (c2352cb)
  • refactor: restructure frontend components, split hooks, add vitest (d770273)
  • refactor: extract detection service, add enum types, DI strategy pattern (c5e221e)
  • fix: E2E detection test selector - text=cat matched multiple elements (dc291e0)
  • chore: add prettier, format codebase, update pre-commit (b9beb11)
  • chore: auto-fix lint in pre-commit hook (b4d60e3)
  • chore: add husky hooks (pre-commit + commit-msg) (8646e8b)
  • chore: remove unused main-page.png (15dd085)
  • docs: add English architecture diagram document (7f6a87e)
  • docs: add Mermaid architecture & workflow diagrams (544f325)
  • fix: swap EN/ZH architecture diagram files (5e15042)
  • docs: add architecture diagram (EN) and workflow diagram (ZH) (bef7e85)
  • docs: add architecture and workflow diagrams (03a95b6)
  • ci: opt into Node.js 24 for GitHub Actions (f4776b9)

Installation

Docker (Recommended)

curl -O https://raw.githubusercontent.com/Somnusochi/VLM-AutoYOLO/master/docker-compose.yml
docker compose up -d

Access:

Docker Images

docker pull ghcr.io/Somnusochi/VLM-AutoYOLO-frontend:v1.5.7
docker pull ghcr.io/Somnusochi/VLM-AutoYOLO-backend:v1.5.7

Manual Setup

See README for detailed instructions.

Full Changelog: v1.5.6...v1.5.7

Assets 4
Loading