Multiple SID robustness..

Samo Penic

2018-11-17 e0996e35862256114826a6314dc649972972a60c

Multiple SID robustness..

 Ocr.py

@@ -10,10 +10,10 @@
        self.filename = filename
        self.invalid = None
        self.QRData = None
        self.settings={'answer_treshold':0.25,} if settings is None else settings
        self.settings = {"answer_threshold": 0.25} if settings is None else settings
        self.errors = []
        self.warnings = []
        self.sid=None
        self.sid = None
        self.sid_classifier = sid_classifier
        if filename is not None:
            self.loadImage(filename)
@@ -57,6 +57,11 @@
            self.data = None
            self.invalid = True
            return
        if(len(d)>1): #if there are multiple codes, get first ean or qr code available.
            for dd in d:
                if(dd.type=="EAN13" or dd.type=="QR"):
                    d[0]=dd
                    break
        self.QRDecode = d
        self.QRData = d[0].data
        xpos = d[0].rect.left
@@ -66,11 +71,11 @@
            self.rotateAngle(180)

    def rotateAngle(self, angle=0):
        #rot_mat = cv2.getRotationMatrix2D(
        # rot_mat = cv2.getRotationMatrix2D(
        #    (self.imgHeight / 2, self.imgWidth / 2), angle, 1.0
        #)
        # )
        rot_mat = cv2.getRotationMatrix2D(
            (self.imgWidth/2, self.imgHeight/2), angle, 1.0
            (self.imgWidth / 2, self.imgHeight / 2), angle, 1.0
        )
        result = cv2.warpAffine(
            self.img,
@@ -222,58 +227,68 @@
        if self.sid_classifier is None:
            return "x"
        if self.settings is not None:
            sid_mask=self.settings.get("sid_mask", None)
        es,err,warn = getSID(
            sid_mask = self.settings.get("sid_mask", None)
        es, err, warn = getSID(
            self.img[
                int(0.04 * self.imgHeight) : int(0.095 * self.imgHeight),
                int(0.7 * self.imgWidth) : int(0.99 * self.imgWidth),
            ],
            self.sid_classifier,
            sid_mask
            sid_mask,
        )
        [self.errors.append(e) for e in err]
        [self.warnings.append(w) for w in warn]
        return es


    def get_code_data(self):
        if self.QRData is None:
            self.errors.append("Could not read QR or EAN code! Not an exam?")
            retval = {'exam_id': None,
                      'page_no': None,
                      'paper_id': None,
                      'faculty_id': None,
                      'sid':None
                      }
            return retval
        qrdata = bytes.decode(self.QRData, 'utf8')
        if self.QRDecode[0].type=='EAN13':
            return {'exam_id': int(qrdata[0:7]),
                    'page_no': int(qrdata[7]),
                    'paper_id': int(qrdata[-5:-1]),
                    'faculty_id': None,
                    'sid': None
                    }
        else:
            data=qrdata.split(',')
            retval={'exam_id': int(data[1]),
                    'page_no': int(data[3]),
                    'paper_id':int(data[2]),
                    'faculty_id':int(data[0]),
            retval = {
                "exam_id": None,
                "page_no": None,
                "paper_id": None,
                "faculty_id": None,
                "sid": None,
            }
            if(len(data)>4):
                retval['sid']=data[4]
            return retval
        qrdata = bytes.decode(self.QRData, "utf8")
        if self.QRDecode[0].type == "EAN13":
            return {
                "exam_id": int(qrdata[0:7]),
                "page_no": int(qrdata[7]),
                "paper_id": int(qrdata[-5:-1]),
                "faculty_id": None,
                "sid": None,
            }
        else:
            data = qrdata.split(",")
            retval = {
                "exam_id": int(data[1]),
                "page_no": int(data[3]),
                "paper_id": int(data[2]),
                "faculty_id": int(data[0]),
            }
            if len(data) > 4:
                retval["sid"] = data[4]

            return retval

    def get_paper_ocr_data(self):
        data=self.get_code_data()
        data['qr']=self.QRData
        data['errors']=self.errors
        data['warnings']=self.warnings
        data['up_position']=(list(self.xMarkerLocations[1]/self.imgWidth), list(self.yMarkerLocations[1]/self.imgHeight))
        data['right_position']=(list(self.xMarkerLocations[1]/self.imgWidth), list(self.yMarkerLocations[1]/self.imgHeight))
        data['ans_matrix']=((np.array(self.answerMatrix)>self.settings['answer_treshold'])*1).tolist()
        if data['sid'] is None:
            data['sid']=self.get_enhanced_sid()
        data = self.get_code_data()
        data["qr"] = self.QRData
        data["errors"] = self.errors
        data["warnings"] = self.warnings
        data["up_position"] = (
            list(self.xMarkerLocations[1] / self.imgWidth),
            list(self.yMarkerLocations[1] / self.imgHeight),
        )
        data["right_position"] = (
            list(self.xMarkerLocations[1] / self.imgWidth),
            list(self.yMarkerLocations[1] / self.imgHeight),
        )
        data["ans_matrix"] = (
            (np.array(self.answerMatrix) > self.settings["answer_threshold"]) * 1
        ).tolist()
        if data["sid"] is None and data["page_no"] == 0:
            data["sid"] = self.get_enhanced_sid()
        return data

 aoiOcr.py

@@ -3,11 +3,11 @@

from glob import glob

settings = {"sid_mask": "64xx0xxx", "answer_treshold": 0.25}
settings = {"sid_mask": "64xx0xxx", "answer_threshold": 0.25}
classifier = joblib.load("filename.joblib")

#p = Paper(filename="testpage300dpi_scan1.png")
#p=Paper(filename='sizif111.tif', sid_classifier=classifier, settings={"sid_mask": "11xx0xxx", "answer_treshold": 0.25})
#p=Paper(filename='sizif111.tif', sid_classifier=classifier, settings={"sid_mask": "11xx0xxx", "answer_threshold": 0.25})
#p=Paper(filename='processed_scans/20141016095134535_0006.tif', sid_classifier=classifier, settings=settings)
#p = Paper(filename="processed_scans/20151111080408825_0001.tif",sid_classifier=classifier,settings=settings,)
#p=Paper(filename='processed_scans/20151028145444607_0028.tif', sid_classifier=classifier, settings=settings)
@@ -20,9 +20,10 @@
    "processed_scans/20141021095744144_0009.tif",
    "processed_scans/20141028095553745_0018.tif",
    "processed_scans/20151013180545275_0011.tif",
    "processed_scans/20160408140801098_0004.tif"
    "processed_scans/20160408140801098_0004.tif",
    "processed_scans/20160510075445995_0026.tif"
]
p=Paper(filename=pa[8], sid_classifier=classifier, settings=settings)
p=Paper(filename=pa[9], sid_classifier=classifier, settings=settings)

# print(p.QRData)
# print(p.errors)
@@ -35,13 +36,20 @@


print(p.get_paper_ocr_data())
exit(0)


filelist = glob("processed_scans/*.tif")
wrong_sid=0;
total=0
for f in sorted(filelist):
    print("processing: {}".format(f))
    print(
        f,
        Paper(
            filename=f, sid_classifier=classifier, settings=settings
        ).get_paper_ocr_data(),
    )
    p=Paper(filename=f, sid_classifier=classifier, settings=settings).get_paper_ocr_data()
    print(f,p)
    if(p['page_no']==0):
        total+=1
    if(len(p['errors'])!=0):
        wrong_sid+=1
    if total%10 == 0:
        print("Total:{}, wrong SID: {}".format(total,wrong_sid))

print("Total:{}, wrong SID: {}".format(total,wrong_sid))

 sid_process.py

@@ -210,11 +210,11 @@
    cv2.imwrite("enhancedSID.png", image)

    sid_no = segment_by_contours(image, image_original, classifier, sid_mask)
    print(sid_no)

    if len(sid_no) != len(sid_mask) or not sid_compare(sid_no, sid_mask):
        sid_warn.append("Trying second SID algorithm.")
        sid_no = segment_by_7segments(image, image_original, sid_mask, classifier)
    print(sid_no)

    if (len(sid_no)) != len(sid_mask):
        sid_no = segment_by_sid_len(image, image_original, sid_mask, classifier)
        sid_warn.append("Trying third SID algorithm.")