{
  "benchmark_id": "imgtotext-ocr-bench-v1",
  "run_date": "2026-08-11",
  "run_at_utc": "2026-08-11T10:31:18.270450+00:00",
  "seed": 20260811,
  "metric": "character_error_rate",
  "normalization": "collapse whitespace (split on any whitespace, rejoin with single spaces, strip)",
  "engines": {
    "gemini": {
      "model": "gemini-2.5-flash",
      "prompt": "same extraction prompt as app/api/ocr/route.js",
      "notes": "Image sent as PNG base64; temperature 0.1"
    },
    "tesseract": {
      "binary": "tesseract 5.5.0",
      "oem": 1,
      "psm": 3,
      "lang": "eng",
      "notes": "System Tesseract LSTM (OEM 1), matching the site's OEM.LSTM_ONLY intent. Site fallback uses tesseract.js in-browser with extra preprocessing; this bench uses the raw PNG with no site preprocessor."
    }
  },
  "corpus_notes": {
    "handwriting_sim": "Not real penmanship \u2014 italic DejaVu Serif with per-glyph jitter on ruled paper. Included to stress irregular spacing, not to claim handwriting CER for humans.",
    "privacy": "All samples are synthetic; no real customer documents."
  },
  "summary": {
    "n_images": 12,
    "mean_cer_tesseract": 0.1975,
    "mean_cer_gemini": 0.0305,
    "by_category": {
      "clean_print": {
        "n": 2,
        "mean_cer_tesseract": 0.0,
        "mean_cer_gemini": 0.0
      },
      "dark_mode_screenshot": {
        "n": 2,
        "mean_cer_tesseract": 0.1477,
        "mean_cer_gemini": 0.1388
      },
      "handwriting_sim": {
        "n": 2,
        "mean_cer_tesseract": 0.0182,
        "mean_cer_gemini": 0.0
      },
      "low_light_photo": {
        "n": 2,
        "mean_cer_tesseract": 0.0042,
        "mean_cer_gemini": 0.0042
      },
      "skewed_scan": {
        "n": 2,
        "mean_cer_tesseract": 0.6487,
        "mean_cer_gemini": 0.0
      },
      "thermal_receipt": {
        "n": 2,
        "mean_cer_tesseract": 0.3662,
        "mean_cer_gemini": 0.0401
      }
    }
  },
  "samples": [
    {
      "id": "clean_print_01",
      "category": "clean_print",
      "file": "corpus/clean_print_01.png",
      "ground_truth_chars": 176,
      "tesseract": {
        "text": "Meeting notes \u2014 11 August 2026 Attendees: Priya, Marcus,\nElena Agenda item 1: ship the receipt OCR guide before\nFriday. Action: Marcus will upload the revised PDF by 17:00\nIST.",
        "cer": 0.0
      },
      "gemini": {
        "text": "Meeting notes \u2014 11 August 2026 Attendees: Priya, Marcus,\nElena Agenda item 1: ship the receipt OCR guide before\nFriday. Action: Marcus will upload the revised PDF by 17:00\nIST.",
        "cer": 0.0,
        "model": "gemini-2.5-flash"
      }
    },
    {
      "id": "clean_print_02",
      "category": "clean_print",
      "file": "corpus/clean_print_02.png",
      "ground_truth_chars": 134,
      "tesseract": {
        "text": "Shipping label Order ID: ORD-88421 Recipient: Jordan Lee\nAddress: 14 Lakeview Road, Pune 411001 Contents: 2x\nUSB-C hub, 1x spare cable",
        "cer": 0.0
      },
      "gemini": {
        "text": "Shipping label Order ID: ORD-88421 Recipient: Jordan Lee\nAddress: 14 Lakeview Road, Pune 411001 Contents: 2x\nUSB-C hub, 1x spare cable",
        "cer": 0.0,
        "model": "gemini-2.5-flash"
      }
    },
    {
      "id": "thermal_receipt_01",
      "category": "thermal_receipt",
      "file": "corpus/thermal_receipt_01.png",
      "ground_truth_chars": 229,
      "tesseract": {
        "text": "CORNER CAFE\nGSTIN 27AABCU9603R1ZM\n\nEspresso 120.00\nMasala dosa 90.00\nBottled water 30.00\nSubtotal 240.00\nCGST 2.5% 6.00\nSGST 2.5% 6.00\nTOTAL 252.00\n\nPaid UPI 12:41 IST\nTxn REF9K2M7Q",
        "cer": 0.2183
      },
      "gemini": {
        "text": "CORNER CAFE\nGSTIN 27AABCU9603R1ZM\n--------------------\nEspresso 120.00\nMasala dosa 90.00\nBottled water 30.00\n--------------------\nSubtotal 240.00\nCGST 2.5% 6.00\nSGST 2.5% 6.00\nTOTAL 252.00\nPaid UPI 12:41 IST\nTxn REF9K2M7Q",
        "cer": 0.0349,
        "model": "gemini-2.5-flash"
      }
    },
    {
      "id": "thermal_receipt_02",
      "category": "thermal_receipt",
      "file": "corpus/thermal_receipt_02.png",
      "ground_truth_chars": 177,
      "tesseract": {
        "text": "CITY MART\nBill No 441902\n\nMilk 1L\nBread loaf\nEggs dozen\n\nItems 3\nTOTAL\n\nCash tendered\nChange\n\n175.\n200.\n25.",
        "cer": 0.5141
      },
      "gemini": {
        "text": "CITY MART\nBill No 441902\n--------------------\nMilk 1L 58.00\nBread loaf 45.00\nEggs dozen 72.00\n--------------------\nItems 3\nTOTAL 175.00\nCash tendered 200.00\nChange 25.00",
        "cer": 0.0452,
        "model": "gemini-2.5-flash"
      }
    },
    {
      "id": "dark_screenshot_01",
      "category": "dark_mode_screenshot",
      "file": "corpus/dark_screenshot_01.png",
      "ground_truth_chars": 169,
      "tesseract": {
        "text": "Terminal \u2014 session\n\n$ mpm run build\n\n> next build\n\nError: ENOENT: no such file or directory\nat Object.openSync (node: fs:596:3)\nat readFileSync (node: fs:464:35)\n\nhint: check app/sitemap.js path",
        "cer": 0.1302
      },
      "gemini": {
        "text": "Terminal \u2014 session\n\n$ npm run build\n> next build\nError: ENOENT: no such file or directory\n    at Object.openSync (node:fs:596:3)\n    at readFileSync (node:fs:464:35)\nhint: check app/sitemap.js path",
        "cer": 0.1124,
        "model": "gemini-2.5-flash"
      }
    },
    {
      "id": "dark_screenshot_02",
      "category": "dark_mode_screenshot",
      "file": "corpus/dark_screenshot_02.png",
      "ground_truth_chars": 115,
      "tesseract": {
        "text": "Terminal \u2014 session\n\nPOST /api/ocr 429\n\ncode: RATE_LIMITED\n\nremaining: 0\n\nresetAt: 2026-08-12T00:00:00Z\nmessage: Daily AI OCR limit reached",
        "cer": 0.1652
      },
      "gemini": {
        "text": "Terminal \u2014 session\n\nPOST /api/ocr 429\ncode: RATE_LIMITED\nremaining: 0\nresetAt: 2026-08-12T00:00:00Z\nmessage: Daily AI OCR limit reached",
        "cer": 0.1652,
        "model": "gemini-2.5-flash"
      }
    },
    {
      "id": "skewed_scan_01",
      "category": "skewed_scan",
      "file": "corpus/skewed_scan_01.png",
      "ground_truth_chars": 121,
      "tesseract": {
        "text": "Library Checkout Slip Member: A Raman Title: Pra\nImaging Due date: 25 August 2026 Barcode: 39\n\nCtical Document\n915012345678",
        "cer": 0.2975
      },
      "gemini": {
        "text": "Library checkout slip Member: A. Raman Title: Practical Document\nImaging Due date: 25 August 2026 Barcode: 39015012345678",
        "cer": 0.0,
        "model": "gemini-2.5-flash"
      }
    },
    {
      "id": "skewed_scan_02",
      "category": "skewed_scan",
      "file": "corpus/skewed_scan_02.png",
      "ground_truth_chars": 125,
      "tesseract": {
        "text": "",
        "cer": 1.0
      },
      "gemini": {
        "text": "Permission slip Student: Kavya Nair Event: Science fair visit Date:\n19 August 2026 Parent signature required before boarding.",
        "cer": 0.0,
        "model": "gemini-2.5-flash"
      }
    },
    {
      "id": "low_light_01",
      "category": "low_light_photo",
      "file": "corpus/low_light_01.png",
      "ground_truth_chars": 121,
      "tesseract": {
        "text": "Whiteboard reminder Deploy window: Saturday 02:00-04:00\nIST Owner: platform team Rollback plan reviewed \u2014 ticket\nPLAT-992",
        "cer": 0.0083
      },
      "gemini": {
        "text": "Whiteboard reminder Deploy window: Saturday 02:00-04:00\nIST Owner: platform team Rollback plan reviewed \u2014 ticket\nPLAT-992",
        "cer": 0.0083,
        "model": "gemini-2.5-flash"
      }
    },
    {
      "id": "low_light_02",
      "category": "low_light_photo",
      "file": "corpus/low_light_02.png",
      "ground_truth_chars": 96,
      "tesseract": {
        "text": "Parking notice Bay B-14 reserved for visitors Valid 11 Aug 2026\nonly Call desk on extension 4400",
        "cer": 0.0
      },
      "gemini": {
        "text": "Parking notice Bay B-14 reserved for visitors Valid 11 Aug 2026\nonly Call desk on extension 4400",
        "cer": 0.0,
        "model": "gemini-2.5-flash"
      }
    },
    {
      "id": "handwriting_sim_01",
      "category": "handwriting_sim",
      "file": "corpus/handwriting_sim_01.png",
      "ground_truth_chars": 55,
      "tesseract": {
        "text": "Buy milk and eggs\nCall dentist Tues day\nFinjsh OCR notes",
        "cer": 0.0364
      },
      "gemini": {
        "text": "Buy milk and eggs\nCall dentist Tuesday\nFinish OCR notes",
        "cer": 0.0,
        "model": "gemini-2.5-flash"
      }
    },
    {
      "id": "handwriting_sim_02",
      "category": "handwriting_sim",
      "file": "corpus/handwriting_sim_02.png",
      "ground_truth_chars": 64,
      "tesseract": {
        "text": "Password reset code\nis four eight two one\nexpires in ten minutes",
        "cer": 0.0
      },
      "gemini": {
        "text": "Password reset code\nis four eight two one\nexpires in ten minutes",
        "cer": 0.0,
        "model": "gemini-2.5-flash"
      }
    }
  ]
}