onnx推理时报错RuntimeError: Input must be a list of dictionaries or a single numpy array for input ‘image

onnx推理时报错：

RuntimeError: Input must be a list of dictionaries or a single numpy array for input 'images'

找到原因啦，onnx输入的字典对应的值应该是numpy类型， 我的代码中是tensor类型，

torch.Size([1, 3, 320, 320]) {'images': tensor([[[[-0.9451, -0.9529, -0.9373,  ..., -0.1059, -0.1059, -0.1059],
          [-0.9373, -0.9373, -0.9294,  ..., -0.1059, -0.1059, -0.1059],
          [-0.9451, -0.9294, -0.9373,  ..., -0.1059, -0.1059, -0.1059],

输入数据

self.face_detect_img = torch.from_numpy(padded_img).unsqueeze(0)

修改为

self.face_detect_img = padded_img[np.newaxis, :]

即可，问题解决