如何使用opencv python将二维numpy数组转换为轮廓

Question

i am trying to convert a 2d array into contours so that i can use drawcounter function of opencv .so if is there any way to convert 2d array into contours then please let suggest me any solution.

2d array:

[[ 5.961938   -0.39692658 -0.02718444 ... -0.50327307  0.05710772
  -0.07550404]
 [ 5.9570045  -0.39401022 -0.0331709  ... -0.5008669   0.04731249
  -0.07948248]
 [ 5.9526873  -0.3914584  -0.03840905 ... -0.49876142  0.03874166
  -0.08296361]
 ...
 [ 6.7853484  -0.6472809  -0.22270808 ... -0.44317752 -0.18152599
  -0.35107702]
 [ 6.794132   -0.65010595 -0.22126621 ... -0.44905755 -0.17756268
  -0.3512323 ]
 [ 6.8041706  -0.6533345  -0.21961835 ... -0.4557776  -0.17303322
  -0.35140973]]

now i want to convert this array into contours .

let me show the code so that it will help to understand . basically i am using self-human-parsing method on pre-trained model. now i want to extract the contours of all classes of outputs.

output = model(image.cuda())
upsample = torch.nn.Upsample(size=input_size, mode='bilinear', 
              align_corners=True)
upsample_output = upsample(output[0][-1][0].unsqueeze(0))
upsample_output = upsample_output.squeeze()
upsample_output = upsample_output.permute(1, 2, 0)  # CHW -> HWC

logits_result = transform_logits(upsample_output.data.cpu().numpy(), c, s, w, h, input_size=input_size)
parsing_result = np.argmax(logits_result, axis=2)

i had used this code from github and i don't have any idea that what they are trying to do. i need only the contours of all classes .

Answer 1

I would use reshape function as mentioned in the comments

    my_array = [[ 5.961938   -0.39692658 -0.02718444 ... -0.50327307  0.05710772
  -0.07550404]
 [ 5.9570045  -0.39401022 -0.0331709  ... -0.5008669   0.04731249
  -0.07948248]
 [ 5.9526873  -0.3914584  -0.03840905 ... -0.49876142  0.03874166
  -0.08296361]
 ...
 [ 6.7853484  -0.6472809  -0.22270808 ... -0.44317752 -0.18152599
  -0.35107702]
 [ 6.794132   -0.65010595 -0.22126621 ... -0.44905755 -0.17756268
  -0.3512323 ]
 [ 6.8041706  -0.6533345  -0.21961835 ... -0.4557776  -0.17303322
  -0.35140973]]

    #then create an opencv contour like follows

    contour = np.array(my_array).reshape(len(my_array), 1, 2)

I didnt test the code. Note that contours' shape is (x, 1, 2) where x is length of your array