我正在關(guān)注text-detection-ctpn。該項(xiàng)目的作者resize()在進(jìn)行計(jì)算之前的圖像。如果不調(diào)整圖像大小，內(nèi)存就會(huì)爆炸。我設(shè)法使用此功能將調(diào)整大小的框轉(zhuǎn)換回原始框def transform_boxes(boxes: np.ndarray, h, w, rh, rw): """ Transform back the original coordinate :param boxes: :param h: height of the original :param w: width of the original :param rh: re-sized height :param rw: re-sized height :return: """ z = np.copy(boxes) z[:, 0] = z[:, 0] / rh z[:, 2] = z[:, 2] / rh z[:, 4] = z[:, 4] / rh z[:, 6] = z[:, 6] / rh z[:, 1] = z[:, 1] / rw z[:, 3] = z[:, 3] / rw z[:, 5] = z[:, 5] / rw z[:, 7] = z[:, 7] / rw return z變換誤差在大坐標(biāo)數(shù)上顯著
2 回答
MMTTMM
TA貢獻(xiàn)1869條經(jīng)驗(yàn) 獲得超4個(gè)贊
當(dāng)您嘗試反轉(zhuǎn)框上的調(diào)整大小操作時(shí)，您除以rhandrw但您永遠(yuǎn)不會(huì)乘以hand w。
z[:, 0] = h * z[:, 0] / rh
z[:, 1] = w * z[:, 1] / rw
這解釋了為什么您的錯(cuò)誤會(huì)隨著更大的圖像而變大。
作為旁注，您可以使用 numpy 索引來避免每行重復(fù)四次：
z[:, 0::2] = h * z[:, 0::2] / rh
z[:, 1::2] = w * z[:, 1::2] / rw
喵喵時(shí)光機(jī)
TA貢獻(xiàn)1846條經(jīng)驗(yàn) 獲得超7個(gè)贊
我必須轉(zhuǎn)換坐標(biāo)而不是我正在處理的像素。
坐標(biāo)，而不是像素....
def transform_boxes(boxes: np.ndarray, im):
"""
Transform back the original coordinate
:param boxes:
:param im: The original image
:return:
"""
z = np.copy(boxes)
(height, width, colors) = im.shape
new_h, new_w, img_size = get_float_new_wh(im)
z[:, 0::2] = height * z[:, 0::2] / new_h
z[:, 1::2] = width * z[:, 1::2] / new_w
return z
def get_new_wh(img):
"""
Get only new width and new height
:param img:
:return:
"""
new_h, new_w, img_size = get_float_new_wh(img)
new_h = int(new_h)
new_w = int(new_w)
new_h = new_h if new_h // 16 == 0 else (new_h // 16 + 1) * 16
new_w = new_w if new_w // 16 == 0 else (new_w // 16 + 1) * 16
return new_h, new_w, img_size
添加回答
舉報(bào)
0/150
提交
取消