CODEWITHSUNDEEP
pythonopencv
doudev
doudev

Reputation: 143

python boundingRect with list of points

I got an error when trying to use the boundingRect() function in opencv. What given is a list of points

lists = []
for match in enumerate(matches):
    lists.append(kp2[match.trainIdx].pt)
x,y,w,h = cv2.boundingRect(lists)

TypeError: points is not a numpy array, neither a scalar

P/s: I want to draw a rectangle around the detected object in the image

Any help is appreciate

Edit

The lists before change to np array

[(328.0, 227.0), (372.0, 241.0), (366.0, 229.0)]

and after

[[ 328. 227.] [ 372. 241.] [ 366. 229.]]

Upvotes: 3

Views: 5774

Answers (2)

waldiphil
waldiphil

Reputation: 75

The list of points needs to be passed as a numpy array. Therefore, first convert it, e.g. with

lists = np.array(lists, dtype=np.float32)

The failed assertion 

'error: (-215) npoints >= 0 && (depth == CV_32F || depth == CV_32S) in function cv::pointSetBoundingRect'

is triggered if the created numpy array does not have the datatype float32 or int32. Another reason would be if the passed array does not contain any points.

The parameter dtype=np.float32 in the array creation command ensures that the created array has the datatype float32. In a 64bit environment, omitting this parameter will lead to the array being created with datatype float64, which causes the assertion to fail.

After the conversion to a numpy array with datatype float32, passing the result to cv.boundingRectangle will work as expected.

Intentionally answering an old question after I was directed here by a search engine.

Upvotes: 2

pancur
pancur

Reputation: 31

I just had exactly the same issue. Didn't resolve it but here is the code that solved the problem:

points = ([1,1], [3,1], [3,3], [1,4])

xx, yy = zip(*points)
min_x = min(xx); min_y = min(yy); max_x = max(xx); max_y = max(yy)
bbox = [(min_x, min_y), (max_x, min_y), (max_x, max_y), (min_x, max_y)]

Upvotes: 3

Related Questions