Skip to main content

OCR Recognition

tip
  • The OCR module performs text recognition on images
  • The OCR module uses the ocr prefix, e.g. ocr.initOcr()
  • Current OCR engines include appleVision
  • 3.18.0+ adds ocrMut as the multi-instance OCR prefix
  • 5.12.0 adds paddleLiteOcr
  • 5.21.0 adds paddleNcnnOcrV5
  • Adds paddleOnnxOcrV6 (PP-OCRv6_small, coexists with paddleOnnxOcr / PP-OCRv5)
  • Languages: both paddleOnnxOcr (v5) and paddleOnnxOcrV6 (v6 Medium/Small) can recognize Simplified/Traditional Chinese, English, Japanese, and many Latin-script languages in a single model — usually you do not need to switch models by language (v6 officially supports ~50 languages; v5 also covers many Latin scripts in practice). Cyrillic scripts (e.g. Ukrainian) are not supported and cannot be recognized correctly
  • Speed: paddleOnnxOcrV6 defaults to accuracy-oriented settings (maxSideLen 960, angle detection enabled). For speed, use the "fast" parameters below (lower maxSideLen, disable doAngle, increase thread count) — usually much faster; enable angle detection only when the image may be upside down

Single-Instance Mode​

ocr.initOcr Initialize​

  • Initialize the OCR module
  • @param map map parameters:
  • keys:
  • type: OCR type — appleVision = iOS built-in Vision module
  • For appleVision, set parameters to: {"type":"appleVision","level":"fast","languages":"zh-Hans,en-US"}
    • level: fast = fast, accurate = accurate
    • languages: recognition languages; default is zh-Hans,en-US (Simplified Chinese and English)
    • Supported: ["en-US", "fr-FR", "it-IT", "de-DE", "es-ES", "pt-BR", "zh-Hans", "zh-Hant"]
  • @return {bool} boolean — success or failure
  • appleVision OCR example
function main() {
 let appleVision = {"type": "appleVision", "level": "accurate", "languages": "zh-Hans,en-US"}
 let inited = ocr.initOcr(appleVision)
 logd("Init result -" + inited);
 if (!inited) {
 loge("error : " + ocr.getErrorMsg());
 return;
 }
 for (var ix = 0; ix < 20; ix++) {
 // Read a bitmap
 let img = image.captureFullScreen();
 if (img == null || img == undefined || img.uuid == null || img.uuid == undefined || img.uuid == "") {
 loge("Failed to read image");
 continue;
 }
 console.time("1")
 logd("start---ocr");
 // Recognize the image
 let result = ocr.ocrImage(img, 20 * 1000, {});
 logd(result)
 if (result) {
 logd("OCR result -> " + JSON.stringify(result));
 for (var i = 0; i < result.length; i++) {
 var value = result[i];
 logd("Text : " + value.label + " x: " + value.x + " y: " + value.y + " width: " + value.width + " height: " + value.height);
 }
 } else {
 logw("No result recognized");
 }

 logd("Elapsed: " + console.timeEnd(1) + " ms")
 image.recycle(img)
 sleep(1000);
 logd("ix = " + ix)
 }
 // Release all resources
 ocr.releaseAll();
}

main();

ocr.ocrImage Recognize Text​

  • Perform OCR on an AutoImage; returns JSON data similar to:
 [
 {
 "label": "奇趣装扮三阶盘化",
 "confidence": 0.48334712,
 "x": 11,
 "y": 25,
 "width": 100,
 "height": 100
 }
]
  • label: recognized text
  • confidence: recognition confidence
  • x: X start coordinate
  • Y: Y start coordinate
  • width: width
  • height: height
  • @param bitmap image
  • @param timeout timeout in milliseconds
  • @param extra extra parameters as a map, e.g. {"token":"xxx"}
  • @return {JSON} JSON object
See common code examples
OCR initialization

ocr.getErrorMsg Get Error Message​

  • Get OCR error message
  • @return {string} null means no error
See common code examples
OCR initialization

ocr.releaseAll Release OCR Resources​

  • Release OCR resources
  • @return {bool} success or failure
See common code examples
OCR initialization

Multi-Instance Mode​

ocrMut.releaseAll Release All​

  • EC standalone 5.17.0+
// See code examples

ocrMut.initOcr Initialize​

  • Initialize the OCR module

  • @param map map parameters:

  • keys:

  • type: OCR type — appleVision = iOS built-in Vision module, tess = TesseractOcr, ocrLite = ncnn neural network ocrLite, paddleOcrOnline = EC bundled PC-side PaddleOCR service, paddleLiteOcr = PaddleLite, paddleNcnnOcrV5 = ncnn PaddleOCR, paddleOnnxOcr = onnxruntime PP-OCRv5 implementation, paddleOnnxOcrV6 = onnxruntime PP-OCRv6_small implementation (coexists with v5)

  • For appleVision, set parameters to: {"type":"appleVision","level":"fast","languages":"zh-Hans,en-US"}

    • level: fast = fast, accurate = accurate
    • languages: recognition languages; default is zh-Hans,en-US (Simplified Chinese and English)
    • Supported: ["en-US", "fr-FR", "it-IT", "de-DE", "es-ES", "pt-BR", "zh-Hans", "zh-Hant"]
      paddleLiteOcr = PaddleLite, paddleOnnxOcr = onnxruntime PP-OCRv5, paddleOnnxOcrV6 = PP-OCRv6_small

  • For paddleLiteOcr / paddleOnnxOcr / paddleOnnxOcrV6:

    • Note: since 5.16+, due to onnxruntime and PaddleLite conflicts, PaddleLite was reimplemented with onnx
    • Example parameters: {"type":"paddleOnnxOcr","cpuThreadNum":2} or {"type":"paddleOnnxOcrV6","cpuThreadNum":2}
    • Languages (paddleOnnxOcr / paddleOnnxOcrV6): a single model recognizes Simplified/Traditional Chinese, English, Japanese, and many Latin-script languages — usually you do not need to switch models by language. v6 Medium/Small officially supports ~50 languages (including 46 Latin-script languages); v5 also recognizes many Latin scripts in practice. Does does not support Cyrillic scripts (e.g. Ukrainian) — cannot recognize correctly
    • Fast configuration (recommended for daily UI screenshots): faster than defaults, e.g. {"type":"paddleOnnxOcrV6","cpuThreadNum":-2,"maxSideLen":640,"doAngle":0,"mostAngle":0,"padding":10}
      • maxSideLen:640: lower detection resolution (default 960) — usually the biggest speed gain; small text may be missed slightly
      • doAngle:0 / mostAngle:0: disable orientation detection and angle voting; set to 1 only when the image may be upside down (~90°–270°)
      • cpuThreadNum:-2 or -1: use more CPU threads (default is often 2)
    • cpuThreadNum: CPU thread count; omit if unsure — -1 = all CPUs, -2 = half of CPUs; adjust for recognition speed
    • modelPath: model path; external path e.g. /sdcard/models/ means under sdcard; built-in models are used by default — omit this field
    • labelPath: training label file path; can be external e.g. /sdcard/labels/ppocr_keys_v1.txt; built-in models are used by default — omit this field (v6 exports dictionary from rec yml in Bundle when built-in; usually no manual file needed)
    • detModelFilename: detection model filename (.onnx), placed under modelPath; built-in models used by default — omit this field
    • recModelFilename: recognition model filename (.onnx), placed under modelPath; built-in models used by default — omit this field
    • clsModelFilename: classification model filename (.onnx), placed under modelPath; built-in models used by default — omit this field (v6 angle classification reuses v5 cls model)
    • padding: white border around image to improve recognition; increase when text boxes do not fully enclose text. Default 10. Can affect recognition speed
    • boxThresh: threshold separating text from background; higher values shrink the text region. Range [0, 1], default 0.3
    • boxScoreThresh: threshold for keeping detected text boxes; higher values mean lower recall. Range [0, 1], default 0.5
    • unClipRatio: controls detected text box size; larger values produce bigger boxes. Range [1.6, 2.0], default 1.6
    • doAngle: enable (1) / disable (0) text orientation detection; needed only for upside-down images (rotated 90°–270°). Default 1
    • mostAngle: enable (1) / disable (0) angle voting (recognize entire image in the most likely text orientation); has no effect when orientation detection is disabled. Default 1
    • maxSideLen: if the image's longest side exceeds max_side_len, scale down proportionally to max_side_len. Default 960; adjust for speed (640 for fast mode)

  • For paddleNcnnOcrV5, parameters:

    • numThread: thread count — -1 = all, -2 = half of device CPUs, 0 = not set; adjust for recognition speed
    • modelsDir: model directory path; omit to use built-in models
    • padding: white border around image to improve recognition; increase when text boxes do not fully enclose text. Default 32; can affect speed
    • maxSideLen: if the image's longest side exceeds max_side_len, scale down proportionally. Default 640; adjust for speed
    • keysName: training label filename; can be external e.g. keys.txt; built-in models used by default — omit this field
    • detName: detection model filename (.param), placed under modelsDir; if named det.param, use det without the .param suffix; built-in models used by default — omit this field
    • recName: recognition model filename (.param), placed under modelsDir; if named rec.param, use rec without the .param suffix; built-in models used by default — omit this field

  • For type tess

    • Set parameters to: {"type":"tess",path:"","rillevel":2,"language":"eng+chi_sim"}
    • path: TesseractOcr traineddata folder path, e.g. /Application/tessdata; best to copy to sandbox with the file module — see example
    • language: TesseractOcr language dataset file to recognize, e.g. chi_sim.traineddata = Simplified Chinese, parameter value is chi_sim; multiple languages joined with +, e.g. chi_sim+eng+num; e.g. eng+chi_sim = TesseractOcr official English and Simplified Chinese; auto-finds traineddata files under path
    • rilLevel: PageIteratorLevel parameter — -1 adaptive, 0: RIL_BLOCK, 1: RIL_PARA, 2: RIL_TEXTLINE, 3: RIL_WORD, 4: RIL_SYMBOL
    • ocrEngineMode: recognition engine type — 0 OEM_TESSERACT_ONLY, 1 OEM_LSTM_ONLY, 2 OEM_TESSERACT_LSTM_COMBINED, 3 OEM_DEFAULT
    • tessedit_char_blacklist: blacklist
    • tessedit_char_whitelist: whitelist

  • For type ocrLite, parameters:

 {
 "type": "ocrLite",
 "padding": 10,
 "maxSideLen": 0,
 "boxScoreThresh": 0.6,
 "boxThresh": 0.3,
 "unClipRatio": 1.6,
 "doAngle": 0,
 "mostAngle": 0
}
* numThread: thread count — `-1` = all, `-2` = half of device CPUs, `0` = not set (program decides); `-2` recommended<br/>
* padding: white border around image to improve recognition; increase when text boxes do not fully enclose text. Default 50.<br/>
* maxSideLen: scale by longest image side; larger = slower but more accurate, smaller = faster but less accurate; `0` = no scaling.<br/>
* boxScoreThresh: text box confidence threshold; decrease when text boxes do not fully enclose text<br/>
* boxThresh: same as above; tune experimentally.<br/>
* unClipRatio: single text box size multiplier; larger values produce bigger boxes.<br/>
* doAngle: enable (1) / disable (0) text orientation detection; needed only for upside-down images (rotated 90°–270°). Default off.<br/>
* mostAngle: enable (1) / disable (0) angle voting; has no effect when orientation detection is disabled. Default off.<br/>
  • For type paddleOcrOnline, download EasyClick-PaddleOcr.zip from cloud storage, extract and run
 {
 "type": "paddleOcrOnline",
 "ocrType": "ONNX_PPOCR_V3",
 "padding": 50,
 "maxSideLen": 0,
 "boxScoreThresh": 0.5,
 "boxThresh": 0.3,
 "unClipRatio": 1.6,
 "doAngleFlag": 0,
 "mostAngleFlag": 0
}
* ocrType: model — ONNX_PPOCR_V3, ONNX_PPOCR_V4, NCNN_PPOCR_V3
* serverUrl: Paddle OCR server address; can deploy on another PC and connect from control center, e.g. 192.168.2.8:9022; change IP when deployed on PC; port 9022 is optional
* padding: white border around image to improve recognition; increase when text boxes do not fully enclose text. Default 50.<br/>
* maxSideLen: scale by longest image side; larger = slower but more accurate, smaller = faster but less accurate; `0` = no scaling.<br/>
* boxScoreThresh: text box confidence threshold; decrease when text boxes do not fully enclose text<br/>
* boxThresh: same as above; tune experimentally.<br/>
* unClipRatio: single text box size multiplier; larger values produce bigger boxes.<br/>
* doAngleFlag: enable (1) / disable (0) text orientation detection; needed only for upside-down images (rotated 90°–270°). Default off.<br/>
* mostAngleFlag: enable (1) / disable (0) angle voting; has no effect when orientation detection is disabled. Default off.<br/>
* limit: OCR requests per second; default 1000. Lower to reduce CPU usage<br/>
* checkImage: verify data is an image (1 yes, 0 no); default off.<br/>
  • @return {bool} boolean — success or failure
function main() {
 // Release all OCR resources at start to avoid leaks from previous runs
 ocrMut.releaseAll();
 logd("Start script...")

 // Initialize an instance
 let ocrtest = ocrMut.newOcr();
 let vision = {"type": "appleVision", "level": "accurate", "languages": "zh-Hans,en-US"}
 // paddleOcr parameters
 let paddleOcrOnline = {
 "type": "paddleOcrOnline",
 "ocrType": "ONNX_PPOCR_V3",
 "serverUrl": "192.168.2.13:9022",
 "limit": 12,
 "checkImage": "1",
 "padding": 200
 }
 let ocrLite = {"type": "ocrLite","numThread":2}
 let paddleLiteOcrMap = {"type": "paddleLiteOcr","cpuThreadNum":2,"cpuPowerMode":"LITE_POWER_FULL"}
 let inited = ocrtest.initOcr(ocrLite)
 // let inited = ocrtest.initOcr(paddleLiteOcrMap)
 if (!inited) {
 loge("inited ocr error : " + ocrtest.getErrorMsg())
 return
 } else {
 logd("ocr inited ok")
 }
 for (let i = 0; i < 3; i++) {
 let img = image.captureFullScreen()
 let ocrResult = ocrtest.ocrImage(img, 20000, null)
 logd("ocrResult " + JSON.stringify(ocrResult));
 if (ocrResult) {
 logd("OCR result -> " + JSON.stringify(ocrResult));
 for (var j = 0; j < ocrResult.length; j++) {
 var value = ocrResult[j];
 logd("Text : " + value.label + " x: " + value.x + " y: " + value.y + " width: " + value.width + " height: " + value.height);
 }
 } else {
 logw("No result recognized");
 }
 image.recycle(img)

 sleep(2000)
 }
 // Release when script finishes; no need to release after every use
 ocrtest.releaseAll()
}

main();

TesseractOcr Example​

function main() {
 // Release all OCR resources at start to avoid leaks from previous runs
 ocrMut.releaseAll();
 logd("Start")
 let ts = file.getSandBoxDir()
 let tessdataDir = ts + "/tessdata"
 logd("tessdataDir=> ", tessdataDir)
 file.mkdirs(tessdataDir)
 // Save traineddata datasets to res folder; auto-copied to tessdataDir
 // Copy English dataset
 let saved = saveResToFile("eng.traineddata", tessdataDir + "/eng.traineddata")
 // Copy Chinese dataset
 let saved2 = saveResToFile("chi_sim.traineddata", tessdataDir + "/chi_sim.traineddata")
 logd("saved ", saved)
 if (!saved) {
 logd("copy eng error")
 return;
 }
 if (!saved2) {
 logd("copy chi_sim error")
 return;
 }
 let tessocr = ocrMut.newOcr()
 // Initialize Chinese + English
 let intx = tessocr.initOcr({
 "type": "ocrLite",
 "path": tessdataDir,
 "language": "eng+chi_sim",
 "rilLevel": 2,
 })

 logd("tessocr initOcr " + intx)
 if (!intx) {
 logd(tessocr.getErrorMsg());
 return
 }

 for (let i = 0; i < 10; i++) {
 // Can use screenshot here instead
 //let aa = image.captureFullScreen();
 let aa = readResAutoImage("2.png")

 console.time(1)
 let rse = tessocr.ocrImage(aa, 10 * 1000, {})
 logd("Elapsed time - ", console.timeEnd(1))
 image.recycle(aa)

 logd(JSON.stringify(rse));
 for (let i = 0; i < rse.length; i++) {
 let a = rse[i]
 logd(JSON.stringify(a))
 let b = a.x + "," + a.y + "," + (a.x + a.width) + "," + (a.y + a.height)
 logd(a.label, b)
 }

 }

 tessocr.releaseAll()

 logd("end--")
}


main();

PaddleOnnxOcr Example​

function main() {
 // Release all OCR resources at start to avoid leaks from previous runs
 ocrMut.releaseAll();
 logd("Start script...")
 let ocrtest = ocrMut.newOcr()
 let modelPath = file.getSandBoxFilePath("")
 let labelPath = file.getSandBoxFilePath("ppocrv5_mobile_labels.txt")
 let detModelFilename = "ch_PP-OCRv5_mobile_det.onnx"
 let recModelFilename = "ch_PP-OCRv5_rec_mobile_infer.onnx"
 let clsModelFilename = "ch_ppocr_mobile_v2.0_cls_infer.onnx"
 // Built-in OCR model
 let paddleOnnxOcrMap1 = {"type": "paddleOnnxOcr", "cpuThreadNum": 2, "padding": 10, "maxSideLen": 960}

 // External model
 let paddleOnnxOcrMap2 = {
 "type": "paddleOnnxOcr", "cpuThreadNum": 2, "padding": 10, "maxSideLen": 960,
 "modelPath": modelPath,
 "labelPath": labelPath,
 "detModelFilename": detModelFilename,
 "recModelFilename": recModelFilename,
 "clsModelFilename": clsModelFilename,
 }
 let inited = ocrtest.initOcr(paddleOnnxOcrMap1)
 if (!inited) {
 loge("inited ocr error : " + ocrtest.getErrorMsg())
 return
 } else {
 logd("ocr inited ok")
 }
 for (let i = 0; i < 100; i++) {
 let img = image.captureFullScreen()


 console.time(1)

 let ocrResult = ocrtest.ocrImage(img, 20000, {"cpuThreadNum": 2, "padding": 10, "maxSideLen": 960})

 logd("ocrResult " + JSON.stringify(ocrResult));
 if (ocrResult) {
 logd("OCR result -> " + JSON.stringify(ocrResult));
 for (var j = 0; j < ocrResult.length; j++) {
 var value = ocrResult[j];
 logd("Text : " + value.label + " " + value.x + "," + value.y + "," + (value.x + value.width) + "," + (value.y + value.height));
 }
 } else {
 logw("No result recognized");
 }
 logd("Elapsed: {} ms", console.timeEnd(1))
 sleep(100)
 image.recycle(img)
 sleep(200)
 }
 // Release when script finishes; no need to release after every use
 ocrtest.releaseAll()
}


main()

PaddleOnnxOcrV6 Example​

  • Use type paddleOnnxOcrV6; built-in PP-OCRv6_small (det/rec ONNX); coexists with paddleOnnxOcr (v5) without conflict
  • Parameter fields same as paddleOnnxOcr; uses App Bundle built-in models when modelPath / filenames are omitted
  • Languages: similar to paddleOnnxOcr (v5) — single model recognizes Simplified/Traditional Chinese, English, Japanese, and many Latin-script languages (v6 Medium/Small officially ~50 languages); Cyrillic scripts like Ukrainian are not supported
  • Fast mode: for upright daily UI screenshots, use maxSideLen:640, doAngle:0, mostAngle:0, cpuThreadNum:-2 (see paddleOnnxOcrV6MapFast below); use defaults / enable angle detection when higher detection accuracy is needed or the image may be upside down
function main() {
 // Release all OCR resources at start to avoid leaks from previous runs
 ocrMut.releaseAll();
 logd("Start script...")
 let ocrtest = ocrMut.newOcr()
 let modelPath = file.getSandBoxFilePath("")
 let labelPath = file.getSandBoxFilePath("ppocrv6_small_labels.txt")
 let detModelFilename = "PP-OCRv6_small_det.onnx"
 let recModelFilename = "PP-OCRv6_small_rec.onnx"
 // v6 angle classification reuses v5 cls
 let clsModelFilename = "ch_ppocr_mobile_v2.0_cls_infer.onnx"

 // Fast mode (recommended for daily UI screenshots): smaller longest side, angle detection off, more CPU
 let paddleOnnxOcrV6MapFast = {
 "type": "paddleOnnxOcrV6",
 "cpuThreadNum": -2,
 "padding": 10,
 "maxSideLen": 640,
 "doAngle": 0,
 "mostAngle": 0
 }

 // Default accuracy-oriented: maxSideLen 960, orientation detection enabled
 let paddleOnnxOcrV6Map1 = {"type": "paddleOnnxOcrV6", "cpuThreadNum": 2, "padding": 10, "maxSideLen": 960}

 // External model (place onnx / dictionary in sandbox yourself)
 let paddleOnnxOcrV6Map2 = {
 "type": "paddleOnnxOcrV6", "cpuThreadNum": 2, "padding": 10, "maxSideLen": 960,
 "modelPath": modelPath,
 "labelPath": labelPath,
 "detModelFilename": detModelFilename,
 "recModelFilename": recModelFilename,
 "clsModelFilename": clsModelFilename,
 }
 let inited = ocrtest.initOcr(paddleOnnxOcrV6MapFast)
 if (!inited) {
 loge("inited ocr error : " + ocrtest.getErrorMsg())
 return
 } else {
 logd("ocr v6 inited ok")
 }
 for (let i = 0; i < 100; i++) {
 let img = image.captureFullScreen()

 console.time(1)

 let ocrResult = ocrtest.ocrImage(img, 20000, {"cpuThreadNum": -2, "padding": 10, "maxSideLen": 640, "doAngle": 0, "mostAngle": 0})

 logd("ocrResult " + JSON.stringify(ocrResult));
 if (ocrResult) {
 logd("OCR result -> " + JSON.stringify(ocrResult));
 for (var j = 0; j < ocrResult.length; j++) {
 var value = ocrResult[j];
 logd("Text : " + value.label + " " + value.x + "," + value.y + "," + (value.x + value.width) + "," + (value.y + value.height));
 }
 } else {
 logw("No result recognized");
 }
 logd("Elapsed: {} ms", console.timeEnd(1))
 sleep(100)
 image.recycle(img)
 sleep(200)
 }
 // Release when script finishes; no need to release after every use
 ocrtest.releaseAll()
}


main()

PaddleNcnnOcrV5 Example​

function main() {
 // Release all OCR resources at start to avoid leaks from previous runs
 ocrMut.releaseAll();
 logd("Start script...")
 let ocrtest = ocrMut.newOcr()
 let modelPath = file.getSandBoxFilePath("")
 let detName = "det"
 let recName = "rec"
 // Built-in OCR model
 let paddleNcnnOcrMap1 = {"type": "paddleNcnnOcrV5", "numThread": 2, "padding": 32, "maxSideLen": 640}

 // External model
 let paddleOnnxOcrMap2 = {
 "type": "paddleNcnnOcrV5", "numThread": 2, "padding": 32, "maxSideLen": 640,
 "modelsDir": modelPath,
 "keysName": "keys.txt",
 "detName": detName,
 "recName": recName
 }
 let inited = ocrtest.initOcr(paddleNcnnOcrMap1)
 if (!inited) {
 loge("inited ocr error : " + ocrtest.getErrorMsg())
 return
 } else {
 logd("ocr inited ok")
 }
 for (let i = 0; i < 100; i++) {
 let img = image.captureFullScreen()


 console.time(1)

 // Dynamic parameters can also be set here
 let ocrResult = ocrtest.ocrImage(img, 20000, {"numThread":2,"padding":32})

 logd("ocrResult " + JSON.stringify(ocrResult));
 if (ocrResult) {
 logd("OCR result -> " + JSON.stringify(ocrResult));
 for (var j = 0; j < ocrResult.length; j++) {
 var value = ocrResult[j];
 logd("Text : " + value.label + " " + value.x + "," + value.y + "," + (value.x + value.width) + "," + (value.y + value.height));
 }
 } else {
 logw("No result recognized");
 }
 logd("Elapsed: {} ms", console.timeEnd(1))
 sleep(100)
 image.recycle(img)
 sleep(200)
 }
 // Release when script finishes; no need to release after every use
 ocrtest.releaseAll()
}


main()

ocrInstance.ocrImage Recognize Text​

  • Perform OCR on an AutoImage; returns JSON data similar to:
 [
 {
 "label": "奇趣装扮三阶盘化",
 "confidence": 0.48334712,
 "x": 11,
 "y": 25,
 "width": 100,
 "height": 100
 }
]
  • label: recognized text
  • confidence: recognition confidence
  • x: X start coordinate
  • Y: Y start coordinate
  • width: width
  • height: height
  • @param bitmap image
  • @param timeout timeout in milliseconds
  • @param extra extra parameters as a map, e.g. {"token":"xxx"}
  • @return {JSON} JSON object
See common code examples
OCR initialization

ocrInstance.getErrorMsg Get Error Message​

  • Get OCR error message
  • @return {string} null means no error
See common code examples
OCR initialization

ocrInstance.releaseAll Release OCR Resources​

  • Release OCR resources
  • @return {bool} success or failure
See common code examples
OCR initialization

PaddleOcrOnline HTTP Calls​

  • For EC below 3.18.0+, OCR can be invoked via HTTP
  • Requires downloading EasyClick-PaddleOcr.zip, extracting and running it
function httpPaddleOcr(filePath) {
 // OCR service address
 // See parameter descriptions above for other options
 let url = "http://192.168.2.13:9022/devapi/uploadOcr"
 let ocrType = "ONNX_PPOCR_V3"
 let limit = "1000"
 let ocrParam = {
 "padding": 50,
 "maxSideLen": 0,
 "boxScoreThresh": 0.5,
 "boxThresh": 0.3,
 "unClipRatio": 1.6,
 "doAngleFlag": 0,
 "mostAngleFlag": 0
 }
 ocrParam = utils.base64Encode(JSON.stringify(ocrParam));
 let param = {
 "ocrType": ocrType,
 "limit": limit,
 "ocrParam": ocrParam
 };
 let files = {
 "file": filePath
 }
 let result = http.httpPost(url, param, files, 20 * 1000, {"User-Agent": "test"});
 if (result == null || result == undefined || result == "") {
 return null;
 }

 try {
 result = JSON.parse(result)
 return result["data"]
 } catch (e) {
 return null;
 }

}

function callPaddleOcrTest() {
 let img = image.captureFullScreen()
 if (!img) {
 loge("Screenshot failed");
 return
 }
 let filePath = file.getSandBoxFilePath("ocrtmp.jpg")
 image.saveTo(img, filePath)
 let result = httpPaddleOcr(filePath);
 logd("result " + JSON.stringify(result));


}

callPaddleOcrTest()