我从形状广播输入数组时遇到错误。通常,这似乎是由于将某个维度为p
的数组转换为某个维度为p+1
或p-1
的数组引起的。但是,我的输入和输出的形状似乎是相同的:3个尺寸。所以我的问题是,我在做什么错了,我该如何解决这个问题?
ValueErrorTraceback (most recent call last)
<ipython-input-21-e821b7e8e0de> in <module>()
108 "overlay_glasses.shape: ", overlay_glasses.shape)
109
--> 110 overlay_img[int(y):int(y+h),int(x):int(x+w)] = overlay_glasses
111
以下是print语句的输出,以便更好地理解上述代码。
'overlay_img.shape: ' , (365, 365, 3),
'overlay_glasses.shape: ', ( 34, 99, 3)
'x: ', 623.26, 'y: ', 265.9
'h: ', 34, 'w: ', 99
较大的代码段:
[omitted code here]
if len(centers)>0:
# change the given value of 2.15 according to the size of the detected face
glasses_width = 2.16*abs(centers[1][0]-centers[0][0])
overlay_img = np.ones(shape = roi_color.shape,
dtype = np.uint8)*255
h,w = glass_img.shape[:2]
scaling_factor = glasses_width/w
overlay_glasses = cv2.resize(src = glass_img,
dsize = None,
fx = scaling_factor, # scale factor along x-axis; when it equals 0, it is computed as \texttt{(double)dsize.width/src.cols}
fy = scaling_factor, # scale factor along y-axis
interpolation = cv2.INTER_AREA) # INTER_AREA: resampling using pixel area relation. It may be a preferred method for image decimation,
x = centers[0][0] if centers[0][0] < centers[1][0] else centers[1][0]
# The x and y variables below depend upon the size of the detected face.
x -= 0.26*overlay_glasses.shape[1]
y += 0.85*overlay_glasses.shape[0]
print("x: ", x,
"y: ", y)
#slice the height, width of the overlay image.
h, w = overlay_glasses.shape[:2]
print("h: ", h,
"w: ", w)
print("overlay_img.shape: ", overlay_img.shape,
"overlay_glasses.shape: ", overlay_glasses.shape)
overlay_img[int(y):int(y+h),int(x):int(x+w)] = overlay_glasses # this is the offending line of code
答案 0 :(得分:1)
您的x
超出了为您提供的overlay_img
创建子图像的范围。图像的尺寸为(365,365,3),但是您将x
设置为623,将x+w
设置为722。这将创建一个空的子图像,无法用{{1}填充}的内容。显然overlay_glasses
坐标有问题。