Hugging Face's logo

Spaces:
phucsd
/
cham-ocr-studio
Running

App Files Community
cham-ocr-studio / Dockerfile

Commit History

  1. fix(generator): add missing PIL ImageFont imports and install git in Dockerfile Running

    phucsd
    Verified signature
    4df8b27

  2. feat(studio): set Cham V24 as default model and temporarily hide V25

    phucsd
    Verified signature
    c81c228

  1. feat(keyboard): optimize Cham virtual keyboard with compact 5-row layout and multi-layout selector

    phucsd
    Verified signature
    d9e8ca5

  2. Upload Dockerfile with huggingface_hub

    phucsd
    Verified signature
    3f98804

  1. feat(studio): deploy academic English UI, 7-tab research methodology, dual-language docs & GitHub integration

    phucsd
    Verified signature
    5c8462b

  2. feat(studio): deploy academic English UI, 7-tab research methodology, dual-language docs & GitHub integration

    phucsd
    Verified signature
    2e53a82

  1. feat(studio): deploy specialized Cham-DBNet detector (F1 99.60%)

    phucsd
    Verified signature
    d521e8d

  1. feat(studio): simplify UI by standardizing on DBNet text segmentation

    phucsd
    Verified signature
    2b1771c

  2. fix(studio): make albumentations optional in generate_data.py and trigger rebuild

    phucsd
    Verified signature
    74c9539

  3. fix(studio): add legacy inference.pdmodel for V24 and trigger rebuild

    phucsd
    Verified signature
    0a99ae6

  4. feat(studio): deploy Cham OCR Model V24 (98.64% acc, stanza prefixes & double danda fixes)

    phucsd
    Verified signature
    b863538

  1. chore: trigger rebuild with latest app.py

    phucsd
    Verified signature
    ff97e30

  2. feat: integrate PaddleOCR DBNet detection into Studio

    phucsd
    Verified signature
    00d1f28

  1. Upload Dockerfile with huggingface_hub

    phucsd
    Verified signature
    e3c573d

  1. Research-backed adaptive clustering: calibrate median_line_height and outlier_gap to prune parchment handle specks [bbox=343..702]

    phucsd
    Verified signature
    cde6afc

  1. Re-assign connected components after true_core_coords deduplication to capture main consonants [ymin=30, ymax=78]

    phucsd
    Verified signature
    2577228

  1. Fix frontend bbox top override to use ymin directly from backend bbox

    phucsd
    Verified signature
    13df67f

  2. Strict background artwork line pruning to eliminate spurious bboxes

    phucsd
    Verified signature
    0ceee51

  3. Add raw meta debug print in do_POST

    phucsd
    Verified signature
    2763a2f

  4. Force Docker cache bust to copy new app.py code

    phucsd
    Verified signature
    61aabfc

  5. Deploy pre-Pass 1 base line deduplication to fix tight bbox [390, 34, 573, 44]

    phucsd
    Verified signature
    fa8b9ed

  6. Add debug do_POST result logging

    phucsd
    Verified signature
    50aa1e8

  7. Verified fix for tight bbox [390, 34, 573, 44] and noise component filtering

    phucsd
    Verified signature
    5d01a5d

  8. Fix cluster filtering logic to prune sky noise specks (bbox: [390, 34, 573, 44])

    phucsd
    Verified signature
    742507f

  1. Deploy 1D Horizontal Clustering and exact grapheme alignment fix

    phucsd
    Verified signature
    1d14134

  2. Implement relative confidence pruning for banner noise lines

    phucsd
    Verified signature
    a54b2be

  3. Tighten NMS iom_thresh to 0.12 for duplicate banner lines

    phucsd
    Verified signature
    a758b5a

  4. Fix non-text background lines with CC filter and 1D NMS

    phucsd
    Verified signature
    3a850ca

  5. Implement non-text line pruning in app.py

    phucsd
    Verified signature
    47e51e8

  6. Fix np.median float casting in app.py

    phucsd
    Verified signature
    4994a6d

  7. Add scikit-image and fix app.py error handling

    phucsd
    Verified signature
    d132451

  8. Upload Dockerfile with huggingface_hub

    phucsd
    Verified signature
    6417130