執行CUDA_VISIBLE_DEVICES=0 python deep_3drecon/test.py 出現error

[jerry4308]

error 訊息: Traceback (most recent call last): File "/home/GeneFace/deep_3drecon/test.py", line 68, in main(0, opt, 'deep_3drecon/datasets/examples') File "/home/GeneFace/deep_3drecon/test.py", line 52, in main im_tensor, lm_tensor = read_data(im_path[i], lm_path[i], lm3d_std) File "/home/GeneFace/deep_3drecon/test.py", line 27, in read_data _, im, lm, _ = align_img(im, lm, lm3d_std) File "/home/GeneFace/deep_3drecon/util/preprocess.py", line 204, in align_img trans_params = np.array([w0, h0, s, t[0], t[1]]) ValueError: setting an array element with a sequence. The requested array has an inhomogeneous shape after 1 dimensions. The detected shape was (5,) + inhomogeneous part.

請問一下，這個問題該如何處理呢? 是我那個步驟有出錯嗎? 謝謝~

[jerry4308]

後來我發現把出錯那行裡面所提的那5個變數，全部轉型成為int資料型態就可以了，已解決!

[Parikshat1998]

@jerry4308 How did you resolve it, can you please explain

[jerry4308]

w0, h0, s, t[0], t[1]

I remember I just changed the type of the above variables as the same type.

^^

[premgadwal]

@jerry4308 can you be more precise

[feifeitu1]

请问是在哪一步转换的呢，为什么我直接转换了还是报这个错？

[skrudrjs]

try this code

trans_params = np.array([w0, h0, s, t[0][0], t[1][0]])

[feifeitu1]

好的，谢谢~

[xiao-keeplearning]

Why this bug author has not fixed so far

[jerry4308]

You can do this to fix the problem.

output1 = s.item()
output2 = t.item(0)
output3 = t.item(1)

trans_params = np.array([w0, h0, output1, output2, output3])

[Pythonpa]

try this code

trans_params = np.array([w0, h0, s, t[0][0], t[1][0]])

YES,this method works very well.