How do I convert a torch tensor to an image to be returned by FastAPI?

Question

I have a torch tensor which I need to convert to a byte object so that I can pass it to starlette's StreamingResponse which will return a reconstructed image from the byte object. I am trying to convert the tensor and return it like so:

def some_unimportant_function(params):
    return_image = io.BytesIO()
    torch.save(some_img, return_image)
    return_image.seek(0)
    return_img = return_image.read()
    
    return StreamingResponse(content=return_img, media_type="image/jpeg")

The below works fine on regular byte objects and my API returns the reconstructed image:

def some_unimportant_function(params):
    image = Image.open(io.BytesIO(some_image))

    return_image = io.BytesIO()
    image.save(return_image, "JPEG")
    return_image.seek(0)
    return StreamingResponse(content=return_image, media_type="image/jpeg")

Using PIL library for this

what am I doing wrong here?

Answer 1

使用torchvision.transforms.ToPILImage()模块将 PyTorch Tensor 转换为 PIL Image 对象，然后将其视为 PIL Image，就像第二个函数一样。这是一个例子。

def some_unimportant_function(params):
    tensor = # read the tensor from disk or whatever
    image = torchvision.transforms.ToPILImage()(tensor.unsqueeze(0))
    return_image = io.BytesIO()
    image.save(return_image, "JPEG")
    return_image.seek(0)
    return StreamingResponse(content=return_image, media_type="image/jpeg")