Join the 1-day Testing & QA Summit featuring 15+ Expert Speakers.Register for FREE! Join TestMu Conference

How to use transform_head method in pandera

Best Python code snippet using pandera_python

Image_Processing_Project.py

Source:Image_Processing_Project.py

1import cv22import numpy as np3import dlib4cap = cv2.VideoCapture('video.mp4')5# cap = cv2.VideoCapture('walking_man.mp4')6detector = dlib.get_frontal_face_detector()7predictor = dlib.shape_predictor("shape_predictor_68_face_landmarks.dat")8#ë³¼ì°ê¸°9def cheek(imgorg):10    img_cheek = imgorg.copy()11    for face in faces:12        landmarks = predictor(imgorg, face)13        #ì¼ë³¼14        x1,x2 = landmarks.part(36).x, landmarks.part(48).x15        y1,y2 = landmarks.part(36).y, landmarks.part(48).y16        x3 = landmarks.part(31).x17        cen_x, cen_y = int((x1+x2)/2), int((y1+y2)/2)18        size = int((x3-cen_x)/2)19        cv2.circle(img_cheek,(cen_x,cen_y),size,(153,70,252),cv2.FILLED)20        #ì¤ë¥¸ë³¼21        x1,x2 = landmarks.part(45).x, landmarks.part(54).x22        y1,y2 = landmarks.part(45).y, landmarks.part(54).y23        right_x, right_y = int((x1 + x2) / 2), int((y1 + y2) / 2)24        cv2.circle(img_cheek, (right_x, right_y), size, (153, 70, 252), cv2.FILLED)25    return img_cheek26def sunglasses(imgorg,sunglass_img):27    img_sun = imgorg.copy()28    point0 = (int(sunglass_img.shape[1] * 0.2), int(sunglass_img.shape[0] * 0.5))29    point2 = (int(sunglass_img.shape[1] * 0.8), int(sunglass_img.shape[0] * 0.5))30    Src = [(point0[0], point0[1]), (point2[0], point2[1])] #ì ê¸ë¼ì¤ ì´ë¯¸ì§ì ê¸°ì¤ì 31    sunglass_gray = cv2.cvtColor(sunglass_img, cv2.COLOR_BGR2GRAY)32    ret, mask = cv2.threshold(sunglass_gray, 170, 255, cv2.THRESH_BINARY_INV)33    mask = cv2.merge((mask, mask, mask))34    mask_inv = cv2.bitwise_not(mask)35    sunglass_mask = cv2.bitwise_and(sunglass_img, mask)  # ì ê¸ë¼ì¤36    eye_mask = cv2.bitwise_and(sunglass_img, mask_inv)  # ìê²½ë§ì¤í¬ì ë37    eye_mask = cv2.bitwise_not(eye_mask)38    for face in faces:39        landmarks = predictor(imgorg, face)40        left_eye_x, left_eye_y = landmarks.part(36).x, landmarks.part(36).y41        right_eye_x, right_eye_y = landmarks.part(45).x, landmarks.part(45).y42        # ì ê¸ë¼ì¤ ì´ë¯¸ì§ì ëìì  ì ì¥43        Dst = [(left_eye_x, left_eye_y), (right_eye_x, right_eye_y)]44        # 2Dë³í íë ¬ ìì±45        ret = cv2.estimateAffinePartial2D(np.array([Src]), np.array([Dst]))46        transform_matrix = ret[0]47        # ì ê¸ë¼ì¤ ìì¹ ì´ë ë³í48        transform_sunglass = cv2.warpAffine(sunglass_mask, transform_matrix, (imgorg.shape[1], img_sun.shape[0]))49        # ë ë§ì¤í¬ ì´ë¯¸ì§ ìì¹ í¬ê¸° ì¡°ì 50        transform_eye = cv2.warpAffine(eye_mask, transform_matrix, (img_sun.shape[1], img_sun.shape[0]))51        sun_face = cv2.bitwise_and(img_sun, transform_eye)52        sun_face = cv2.addWeighted(sun_face, 0.5, transform_sunglass, 0.4, 0)53        face_without_eye = cv2.bitwise_and(img_sun, cv2.bitwise_not(transform_eye))54        img_sun = cv2.add(sun_face, face_without_eye)55    return img_sun56def head(imgorg,hat_img):57    img_head = imgorg.copy()58    point0 = (int(hat_img.shape[1] * 0.1), int(hat_img.shape[0] * 0.5)) #ë¨¸ë¦¬ë  ì´ë¯¸ì§ì ê¸°ì¤ì 59    point2 = (int(hat_img.shape[1] * 0.9), int(hat_img.shape[0] * 0.5))60    Src = [(point0[0]-40, point0[1]), (point2[0], point2[1])]61    head_gray = cv2.cvtColor(hat_img, cv2.COLOR_BGR2GRAY)62    ret, mask = cv2.threshold(head_gray, 180, 255, cv2.THRESH_BINARY_INV)63    mask = cv2.merge((mask, mask, mask))64    mask_inv = cv2.bitwise_not(mask)65    hat_mask = cv2.bitwise_and(hat_img, mask)  # ë¨¸ë¦¬ë 66    head_mask = cv2.bitwise_and(hat_img, mask_inv)  # ë¨¸ë¦¬ë ë§ì¤í¬67    head_mask = cv2.bitwise_not(head_mask)68    for face in faces:69        # ì¼êµ´ìì ì¢í70        landmarks = predictor(imgorg, face)71        x1, y1 = face.left(), face.top()72        x2, y2 = face.right(), face.bottom()73        y3 = int(y1 + ((y1 - y2) * 0.3))74        # ì ê¸ë¼ì¤ ì´ë¯¸ì§ì ëìì  ì ì¥75        Dst = [(x1 - 40, y3), (x2 + 10, y3)]76        # 2Dë³í íë ¬ ìì±77        ret = cv2.estimateAffinePartial2D(np.array([Src]), np.array([Dst]))78        transform_matrix = ret[0]79        # ë¨¸ë¦¬ë  ìì¹ ì´ë ë³í80        transform_hat = cv2.warpAffine(hat_mask, transform_matrix, (imgorg.shape[1], img_head.shape[0]))81        # ë ê²ì¶ ë§ì¤í¬ ì´ë¯¸ì§ ìì¹ í¬ê¸° ì¡°ì 82        transform_head = cv2.warpAffine(head_mask, transform_matrix, (img_head.shape[1], img_head.shape[0]))83        head_face = cv2.bitwise_and(img_head, transform_head)84        head_face = cv2.addWeighted(head_face, 0.0, transform_hat, 1.0, 0)85        face_without_eye = cv2.bitwise_and(img_head, cv2.bitwise_not(transform_head))86        img_hat = cv2.add(head_face, face_without_eye)87    return img_hat88def faceblur(imgorg):89    # ì¼êµ´ ììì ê²ì¶í´ ì¼êµ´ìì íëë¦¬ë¥¼ íìíê³  ê·¸ ë¶ë¶ë§ blur ì²ë¦¬í©ëë¤90    for face in faces:91        x1, y1 = face.left(), face.top()92        x2, y2 = face.right(), face.bottom()93        cv2.rectangle(imgorg, (x1, y1), (x2, y2), (0, 0, 0), 2)94        b_face = imgorg[y1:y2, x1:x2]95        b_face = cv2.blur(b_face, (10, 10))96        imgorg[y1:y2, x1:x2] = b_face97    return imgorg98while True:99    ret, img = cap.read()100    if not ret:101        break102    hat_img = cv2.imread("./images/headpin.png") #ë¨¸ë¦¬ë  ì´ë¯¸ì§103    sunglass_img = cv2.imread("./images/sunglass.png") #ì ê¸ë¼ì¤ ì´ë¯¸ì§104    scaler = 0.3 # ìì ì¬ì´ì¦ì¡°ì 105    img = cv2.resize(img, (int(img.shape[1] * scaler), int(img.shape[0] * scaler)))106    imgorg = img.copy()107    faces = detector(imgorg)108    #í¨ìí¸ì¶109    cheek_face = cheek(imgorg)110    sunglass_face = sunglasses(imgorg,sunglass_img)111    hat_face = head(imgorg, hat_img)112    blur_face = faceblur(imgorg)113    # cv2.imshow("img", img)114    cv2.imshow("cheek_face", cheek_face)115    cv2.imshow("sunglass_face", sunglass_face)116    cv2.imshow("hat_face", hat_face)117    cv2.imshow("blur_img",blur_face)...

learn_pyltp.py

Source:learn_pyltp.py

...68    arcs = get_arcs(words)  # å¥æ³åæ69    #print("\t".join("%d:%s" % (arc.head,arc.relation) for arc in arcs))70    parse = [arc.relation for arc in arcs]71    head = [arc.head for arc in arcs]72    head = transform_head(head)73    return parse,head74    75    76def get_role(words):77    postags = get_postag(words)# è¯æ§æ æ³¨78    arcs = get_arcs(words)  # å¥æ³åæ79    roles = labeller.label(words, postags, arcs)  # è¯ä¹è§è²æ æ³¨80    for role in roles:81        print (role.index, "".join(["%s:(%d,%d)" % \82                                    (arg.name, arg.range.start, arg.range.end)\83                                    for arg in role.arguments]))84    85    labeller.release()  # éæ¾æ¨¡å86def get_recognize(words):87    postags = get_postag(words)88    netags = recognizer.recognize(words, postags)  # å½åå®ä½è¯å«89    return netags90    recognizer.release()  # éæ¾æ¨¡å91def transform_head(head):92    '''93    å ä¸ºltpçè¯æ³å³ç³»æ¯ä»rootèç¹å¼å§çï¼94    0A,1Båæäº0rootï¼1A,2B95    ä¸ºäºåæä½¿ç¨æ¹ä¾¿ï¼æä»¬å¯ä»¥éè¿ä¸ä¸ªè½¬æ¢ï¼è®©è¯æ³å³ç³»ä¸è¯è¡¨çè¯ä¸ä¸å¯¹åº96    '''97    head_new = []98    for i in head:99        if i == 0:100            i=0101        else:102            i = i-1103        head_new.append(i)104    return head_new105if __name__ == '__main__':...

marker_roi_heads.py

Source:marker_roi_heads.py

...31        self.decoder_head = decoder_head32    @classmethod33    def from_config(cls, cfg, input_shape):34        ret = super().from_config(cfg)35        transform_head = build_transform_head(cfg, input_shape)36        ret["transform_head"] = transform_head37        corner_input_shape, decoder_input_shape = transform_head.output_shape38        ret["corner_head"] = build_corner_head(cfg, corner_input_shape)39        ret["decoder_head"] = build_decoder_head(cfg, decoder_input_shape)40        return ret41    @property42    def device(self):43        return self.corner_head.device44    45    def forward(46        self,47        images: ImageList,48        features: Dict[str, torch.Tensor],49        proposals: List[Instances],50        targets: Optional[List[Instances]] = None,51    ) -> Tuple[List[Dict], Dict]:52        # del images53        if self.training:54            assert targets55            proposals = self.label_and_sample_proposals(proposals, targets)56        # del targets57        58        if self.training:59            corner_features, decoding_features, sample_locations_batch, losses = self.transform_head(images, features, proposals, targets)60            losses.update(self.corner_head(corner_features, proposals))61            losses.update(self.decoder_head(decoding_features, proposals))62            del images, targets63            return [], losses64        65        corner_features, decoding_features, sample_locations_batch, _ = self.transform_head(images, features, proposals, targets)66        corners_batch = self.corner_head(corner_features, proposals)67        obj_scores_batch, decoded_messages_batch = self.decoder_head(decoding_features, proposals)68        results = []69        for i in range(len(proposals)):70            output = {71                "corners": corners_batch[i], "obj_scores": obj_scores_batch[i],72                "decoded_messages": decoded_messages_batch[i],73                "image_shape": proposals[i].image_size}74            if sample_locations_batch:75                output["sample_locations"] = sample_locations_batch[i]76            results.append(output)...

Automation Testing Tutorials

Learn to execute automation testing from scratch with LambdaTest Learning Hub. Right from setting up the prerequisites to run your first automation test, to following best practices and diving deeper into advanced test scenarios. LambdaTest Learning Hubs compile a list of step-by-step guides to help you be proficient with different test automation frameworks i.e. Selenium, Cypress, TestNG etc.