def text_detect(img,
MAX_HORIZONTAL_GAP=30,
MIN_V_OVERLAPS=0.6,
MIN_SIZE_SIM=0.6,
TEXT_PROPOSALS_MIN_SCORE=0.7,
TEXT_PROPOSALS_NMS_THRESH=0.3,
TEXT_LINE_NMS_THRESH=0.3):
boxes, scores = detect.text_detect(np.array(img))
boxes = np.array(boxes, dtype=np.float32)
scores = np.array(scores, dtype=np.float32)
textdetector = TextDetector(MAX_HORIZONTAL_GAP, MIN_V_OVERLAPS,
MIN_SIZE_SIM)
shape = img.shape[:2]
boxes = textdetector.detect(boxes, scores[:, np.newaxis], shape,
TEXT_PROPOSALS_MIN_SCORE,
TEXT_PROPOSALS_NMS_THRESH,
TEXT_LINE_NMS_THRESH)
text_recs = get_boxes(boxes)
newBox = []
rx = 1
ry = 1
for box in text_recs:
x1, y1 = (box[0], box[1])
x2, y2 = (box[2], box[3])
x3, y3 = (box[6], box[7])
x4, y4 = (box[4], box[5])
newBox.append([
x1 * rx, y1 * ry, x2 * rx, y2 * ry, x3 * rx, y3 * ry, x4 * rx,
y4 * ry
])
return newBox
def text_detect(img,
MAX_HORIZONTAL_GAP=30,
MIN_V_OVERLAPS=0.6,
MIN_SIZE_SIM=0.6,
TEXT_PROPOSALS_MIN_SCORE=0.7,
TEXT_PROPOSALS_NMS_THRESH=0.3,
TEXT_LINE_NMS_THRESH=0.3,
bili=1.2): #下面8行检测单个文字
#下面几行是用yolo给出框.
Image.fromarray(img).save("look.png") #看看boxes,scores的含义 是所有rpn的结果.
boxes, scores = detect.text_detect(np.array(img)) #这里面用的是yolo
boxes = np.array(boxes, dtype=np.float32)
scores = np.array(scores, dtype=np.float32)
Allboxes = boxes
AllScores = scores
#函数下面部分是做行拼接.
textdetector = TextDetector(MAX_HORIZONTAL_GAP, MIN_V_OVERLAPS,
MIN_SIZE_SIM)
shape = img.shape[:2]
#看看下行boxes 的含义. scores:表示最后抽取的汉字对应的score?????????对的,下行的scores就是最后每行的
#分数了!!!!!!!!!!!!!1 非常重要的参数. #下面几行做文字box拼接成seq #tp_groups 表示每一行的文字对应
#boxesForSingle 中的index
boxes, scores, keepIndForSingle, tp_groups, boxesForSingle, scoresForSingle = textdetector.detect(
boxes, scores[:, np.newaxis], shape, TEXT_PROPOSALS_MIN_SCORE,
TEXT_PROPOSALS_NMS_THRESH, TEXT_LINE_NMS_THRESH, bili)
#tp_groups 是boxes对应的 box标号.
text_recs = get_boxes(boxes)
print(text_recs.shape, "text_recs.shape")
newBox = []
rx = 1
ry = 1
for box in text_recs:
x1, y1 = (box[0], box[1])
x2, y2 = (box[2], box[3])
x3, y3 = (box[6], box[7])
x4, y4 = (box[4], box[5])
newBox.append([
x1 * rx, y1 * ry, x2 * rx, y2 * ry, x3 * rx, y3 * ry, x4 * rx,
y4 * ry
])
return newBox, scores, boxesForSingle, scoresForSingle, keepIndForSingle, tp_groups, Allboxes, AllScores
def box_cluster(img, boxes, scores, **args):
MAX_HORIZONTAL_GAP = args.get('MAX_HORIZONTAL_GAP', 30)
MIN_V_OVERLAPS = args.get('MIN_V_OVERLAPS', 0.6)
MIN_SIZE_SIM = args.get('MIN_SIZE_SIM', 0.6)
textdetector = TextDetector(MAX_HORIZONTAL_GAP, MIN_V_OVERLAPS,
MIN_SIZE_SIM)
shape = img.shape[:2]
TEXT_PROPOSALS_MIN_SCORE = args.get('TEXT_PROPOSALS_MIN_SCORE', 0.07)
TEXT_PROPOSALS_NMS_THRESH = args.get('TEXT_PROPOSALS_NMS_THRESH', 0.7)
TEXT_LINE_NMS_THRESH = args.get('TEXT_LINE_NMS_THRESH', 0.9)
LINE_MIN_SCORE = args.get('LINE_MIN_SCORE', 0.07)
boxes, scores = textdetector.detect(boxes, scores[:, np.newaxis], shape,
TEXT_PROPOSALS_MIN_SCORE,
TEXT_PROPOSALS_NMS_THRESH,
TEXT_LINE_NMS_THRESH, LINE_MIN_SCORE)
return boxes, scores
def net_output_process(batch_preds,
batch_shape,
batch_shape_padded,
prob=0.05):
"""
将主干网络的批输出转换为boxes,scores,这里方便兼容之前代码，
暂且使用for循环,以后可替换为vectorize处理
@params batch_preds(list of arrays):list长度代表n个采样尺度，其中每个\
array形状为(batch_size,grid_size_w,grid_size_h,3*(4+1+num_classes))
@params batch_shape(list of tuples):图片原始长宽
@params prob(float):置信度小于prob的box将被忽略
@returns batch_boxes(array): [字符区域数量，8]
@returns batch_scores:[字符区域数量,]
"""
batch_boxes = []
batch_scores = []
MAX_HORIZONTAL_GAP = 100
MIN_V_OVERLAPS = 0.6
MIN_SIZE_SIM = 0.6
textdetector = TextDetector(MAX_HORIZONTAL_GAP, MIN_V_OVERLAPS,
MIN_SIZE_SIM)
TEXT_PROPOSALS_MIN_SCORE = 0.1
TEXT_PROPOSALS_NMS_THRESH = 0.3
TEXT_LINE_NMS_THRESH = 0.99
LINE_MIN_SCORE = 0.1
leftAdjustAlph = 0.01
rightAdjustAlph = 0.01
# 首先初步对模型主干输出进行预处理
for y1, y2, y3, image_shape, input_shape in zip(batch_preds[0],
batch_preds[1],
batch_preds[2],
batch_shape,
batch_shape_padded):
outputs = [y1, y2, y3, image_shape, input_shape]
box, scores = box_layer(outputs, anchors, num_classes)
h, w = image_shape
keep = np.where(scores > prob)
# box[:, 0:4][box[:, 0:4]<0] = 0
box = np.array(box)
scores = np.array(scores)
box[box < 0] = 0
box[:, 0][box[:, 0] >= w] = w - 1
box[:, 1][box[:, 1] >= h] = h - 1
box[:, 2][box[:, 2] >= w] = w - 1
box[:, 3][box[:, 3] >= h] = h - 1
boxes = box[keep[0]]
scores = scores[keep[0]]
# 筛选出需要的box，并且进行nms，字符行组合
boxes, scores = textdetector.detect(boxes, scores[:, np.newaxis],
(h, w), TEXT_PROPOSALS_MIN_SCORE,
TEXT_PROPOSALS_NMS_THRESH,
TEXT_LINE_NMS_THRESH,
LINE_MIN_SCORE)
boxes = sort_box(boxes)
batch_boxes.append(boxes)
batch_scores.append(scores)
# print('done')
return batch_boxes, batch_scores