import cv2
import random
import numpy as np
import imgaug as ia
from preprocess.flag import flag
import imgaug.augmenters as iaa
from preprocess.labelgen import label_generator
f = flag()
size = f.target_size
def augment(image, bbox):
x = random.randint(-100, 100)
y = random.randint(-100, 100)
aug = iaa.Sequential([iaa.Multiply(random.uniform(0.15, 1.5)),
iaa.AdditiveGaussianNoise(random.uniform(0, 0.1) * 255),
iaa.Affine(translate_px={"x": x, "y": y},
scale=random.uniform(0.35, 1.5),
rotate=random.uniform(-180, 180),
cval=(0, 255))])
bbs = ia.BoundingBoxesOnImage([
ia.BoundingBox(x1=bbox[0], y1=bbox[1], x2=bbox[2], y2=bbox[3])], shape=image.shape)
aug = aug.to_deterministic()
image_aug = aug.augment_image(image)
bbs_aug = aug.augment_bounding_boxes([bbs])[0]
b = bbs_aug.bounding_boxes
bbs_aug = [b[0].x1, b[0].y1, b[0].x2, b[0].y2]
bbs_aug = np.asarray(bbs_aug)
import cv2
import numpy as np
from solo import hand_model
from preprocess.flag import flag
model = hand_model()
model.load_weights('weights/solo.h5')
grid_size = flag().grid_size
cam = cv2.VideoCapture(0)
while True:
ret, ori_image = cam.read()
height, width, _ = ori_image.shape
if ret is False:
break
image = cv2.resize(ori_image, (416, 416))
img = image / 255.0
img = np.expand_dims(img, axis=0)
grid_output = model.predict(img)
grid_output = grid_output[0]
output = (grid_output > 0.6).astype(int)
""" Finding bounding box """
prediction = np.where(output > 0.5)
row_wise = prediction[0]
col_wise = prediction[1]
try:
x1 = min(col_wise) * grid_size
y1 = min(row_wise) * grid_size