def __rnet_detect(self, image, pnet_boxes):
pnet_boxes_ = util.convert_to_square(pnet_boxes) # 转正方形
img_datas = []
for box in pnet_boxes_:
x1_ = int(box[0])
y1_ = int(box[1])
x2_ = int(box[2])
y2_ = int(box[3])
img = image.crop((x1_, y1_, x2_, y2_))
img = img.resize((24, 24), Image.ANTIALIAS)
img_data = self.transform(img)
img_datas.append(img_data)
img_datas = torch.stack(img_datas)
cond, offset = self.r_net(
img_datas) # cond的形式为(batch,1), offset的形式为(batch,4)
cond = cond.detach().numpy()
offset = offset.detach().numpy()
indexs, _ = np.where(cond > 0.7)
boxes = []
for index in indexs:
box = pnet_boxes_[index]
_x1 = int(box[0])
_y1 = int(box[1])
_x2 = int(box[2])
_y2 = int(box[3])
ow = _x2 - _x1
oh = _y2 - _y1
x1 = offset[index][0] * ow + _x1
y1 = offset[index][1] * oh + _y1
x2 = offset[index][2] * ow + _x2
y2 = offset[index][3] * oh + _y2
boxes.append([x1, y1, x2, y2, cond[index][0]])
return util.NMS(np.array(boxes), thresh=0.5)
def __onet_detect(self, image, rnet_boxes):
_img_dataset = []
_rnet_boxes = util.convert_to_square(rnet_boxes)
for _box in _rnet_boxes:
_x1 = int(_box[0])
_y1 = int(_box[1])
_x2 = int(_box[2])
_y2 = int(_box[3])
img = image.crop((_x1, _y1, _x2, _y2))
img = img.resize((48, 48), Image.ANTIALIAS)
img_data = self.transform(img)
_img_dataset.append(img_data)
img_dataset = torch.stack(_img_dataset)
_cls, _offset = self.o_net(img_dataset)
cls = _cls.detach().numpy()
offset = _offset.detach().numpy()
boxes = []
idxs, _ = np.where(cls > 0.97)
for idx in idxs:
_box = _rnet_boxes[idx]
_x1 = int(_box[0])
_y1 = int(_box[1])
_x2 = int(_box[2])
_y2 = int(_box[3])
ow = _x2 - _x1
oh = _y2 - _y1
x1 = _x1 + ow * offset[idx][0]
y1 = _y1 + oh * offset[idx][1]
x2 = _x2 + ow * offset[idx][2]
y2 = _y2 + oh * offset[idx][3]
boxes.append([x1, y1, x2, y2, cls[idx][0]])
# O网络最后使用(交集/最小值)的方法做IOU运算
return util.NMS(np.array(boxes), isMin=True, thresh=0.7)
def __pnet_detect(self, image):
boxes = []
img = image
w, h = img.size
min_side_len = min(w, h) # 最小边长
scale = 1 # 缩放比例
# 通过循环生成图像金字塔
while min_side_len > 12: # P网络建议框长度为12
img_data = self.transform(img) # 向量化
img_data.unsqueeze_(0) # 升维，由 CHW 转为 1CHW,相当于加了个批次
cond, offset = self.p_net(img_data) # 1CHW
cond_ = cond[0][0]
offset_ = offset[0]
cond_mask = cond_ > 0.6 # 置信度大于0.6
indexs = torch.nonzero(cond_mask) # 找出置信度大于0.6的下标
for index in indexs:
orignal_box = self.__restore_box(index, cond_[index[0],
index[1]],
offset_, scale)
boxes.append(orignal_box)
rate = 0.9
scale *= rate
_w = int(w * scale)
_h = int(h * scale)
img = img.resize([_w, _h])
min_side_len = min(_w, _h)
return util.NMS(np.array(boxes), 0.5) # nms操作
for id_, box in enumerate(neg_box):
neg_db_tmp[id_, :] = util.img2array(box[5], param.img_size_12)
calib_result = net_12_calib.prediction.eval(
feed_dict={input_12_node: neg_db_tmp})
neg_box = util.calib_box(neg_box, calib_result, img)
#NMS for each scale
scale_cur = 0
scale_box = []
suppressed = []
for id_, box in enumerate(neg_box):
if box[6] == scale_cur:
scale_box.append(box)
if box[6] != scale_cur or id_ == len(neg_box) - 1:
suppressed += util.NMS(scale_box)
scale_cur = box[6]
scale_box = [box]
neg_box = suppressed
suppressed = []
if sys.argv[1] == str(param.img_size_48):
#24-net
result_db_12 = np.zeros((len(neg_box), param.img_size_12,
param.img_size_12, param.input_channel),
np.float32)
result_db_24 = np.zeros((len(neg_box), param.img_size_24,
param.img_size_24, param.input_channel),
np.float32)
for bid, box in enumerate(neg_box):
for id_, box in enumerate(result_box):
result_db_tmp[id_, :] = util.img2array(box[5], param.img_size_12)
calib_result = net_12_calib.prediction.eval(
feed_dict={input_12_node: result_db_tmp})
result_box = util.calib_box(result_box, calib_result, img)
#NMS for each scale
scale_cur = 0
scale_box = []
suppressed = []
for id_, box in enumerate(result_box):
if box[6] == scale_cur:
scale_box.append(box)
if box[6] != scale_cur or id_ == len(result_box) - 1:
suppressed += util.NMS(scale_box)
scale_cur = box[6]
scale_box = [box]
result_box = suppressed
suppressed = []
#24-net
result_db_12 = np.zeros((len(result_box), param.img_size_12,
param.img_size_12, param.input_channel),
np.float32)
result_db_24 = np.zeros((len(result_box), param.img_size_24,
param.img_size_24, param.input_channel),
np.float32)
for bid, box in enumerate(result_box):
resized_img_12 = util.img2array(box[5], param.img_size_12)
