BentoML

Looks like you're not specifying the content type in of your image in your `MultipartEncoder`?

I printed the m.content_type for the second chunk of code, and it gives me
```multipart/form-data; boundary=b67d737b1cc94a55a76d95da35648314```
Is this what you are suggesting?

No, the content type of the file itself inside the multipart form is not being set.

I guess the content type is sent here in the <http://requests.post|requests.post> function
```   res = <http://requests.post|requests.post>(
        "<http://0.0.0.0:3000/predict>", data=m, headers={"Content-Type": m.content_type}
    )```
If not, can you show me an example of how to specify the content type in the image? For example, my content type is image/jpeg.

```    m = MultipartEncoder(
        fields={
        "image": ('abc.jpg', open("/projects/ff_project/users/ff_li/_nasHome/project/ff-lib/bento/abc.jpg", "rb"), 'image/jpeg'),
        "img_type":('', 'mic'),
        "bbox": ('', '[0,0,100,100]'),
        }
    )```

Oh sorry, I did specify the 'image/jpeg' in my code but I forgot to put it here.  I still have this issue even if the image content type is explicitly specified.

Hm, and the error is the same even with the content type specified? Does it work if you use the swagger UI?

I haven't tried swagger UI before.  Is there any other method that I can try to detect the issue?

Sure, you can use curl or any other request library, if it's still a problem then it might be a bug with BentoML, otherwise it's an issue with the request, somehow or other.

I tried CURL in the terminal shell. It still didn't work. So it might be a bug with BentoML?

```curl -X POST -H "Content-Type: multipart/form-data"  -F image=@/projects/ff_project/users/ff_li/_nasHome/project/ff-lib/bento/abc.jpg -F img_type='mic' -F bbox='[0,0,100,100]' <http://0.0.0.0:3000/predict>```

Thanks! Can you open an issue so we can track this one?