How to convert the uploaded image to Numpy array?

[pavitrashah]

Example Code To Get an Image

from fastapi import FastAPI, UploadFile, File, Form

app = FastAPI()


@app.post("/")
def read_root(file: bytes = File(...)):
    return {"Hello": "World"}

Or

from fastapi import FastAPI, UploadFile, File, Form

app = FastAPI()


@app.post("/")
def read_root(file: UploadFile = File(...)):
    return {"Hello": "World"}

Environment

• OS: Windows 10
• FastAPI Version == 0.61.2
• Python version: 3.7

Question

How to convert the uploaded image to Numpy array? So it can be then used in libraries like openCV, tensorflow for Computer Vision or Deep Learning Applications.

Things I have already tried

from fastapi import FastAPI, UploadFile, File, Form
from PIL import Image
from io import BytesIO
import numpy as np

app = FastAPI()

def read_imagefile(data) -> Image.Image:
    image = Image.open(BytesIO(data))
    return image

@app.post("/")
async def read_root(file: UploadFile = File(...)):
    image = read_imagefile(await file.read())
    return {"Hello": "World"}

And

from fastapi import FastAPI, UploadFile, File, Form
from PIL import Image
from io import BytesIO
import numpy as np

app = FastAPI()

def load_image_into_numpy_array(data):
    return np.array(Image.open(BytesIO(data)))


@app.post("/")
async def read_root(file: UploadFile = File(...)):
    image = load_image_into_numpy_array(await file.read())
    return {"Hello": "World"}

But I keep getting errors. And none of the solutions in any other thread works.
Errors: AttributeError: 'JpegImageFile' object has no attribute 'read' or
AttributeError: 'bytes' object has no attribute 'read' or
AttributeError: 'numpy.ndarray' object has no attribute 'read' or
UnicodeDecodeError: 'utf-8' codec can't decode byte 0xff in position 0: invalid start byte

[includeamin]

both of the codes are working for me.

@app.post("/")
async def read_root(file: UploadFile = File(...)):
    image = load_image_into_numpy_array(await file.read())
    print(image.data)
    return {"Hello": "World"}

curl -X POST "http://localhost:3000/" -H  "accept: application/json" -H  "Content-Type: multipart/form-data" -F "file=@image (31).png;type=image/png"

here the output:

[pavitrashah]

Thanks for the reply. Both the codes also worked out for me apparently I had made a mistake in other part of my code.

[tiangolo]

Thanks for the help here @includeamin ! 👏 🙇

Thanks for reporting back and closing the issue @pavitrashah 👍

[belito3]

Probably, we need convert image's color from BGR to RGB

def load_image_into_numpy_array(data):
    npimg = np.frombuffer(data, np.uint8)
    frame = cv2.imdecode(npimg, cv2.IMREAD_COLOR)
    cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
    return frame

[chaitanyakkumar]

hi guys. I am working on a similar code like yours but instead of printing the numpy array I need to return the numpy array. While returning the numpy array I am getting internal server error

[Noctale]

https://pynative.com/python-serialize-numpy-ndarray-into-json/#:~:text=Use%20the%20cls%20kwarg%20of,array%20into%20JSON%20formatted%20data.&text=To%20serialize%20Numpy%20array%20into%20JSON%20we%20need%20to%20convert,using%20a%20tolist()%20function.

this work for me to return numpy (but it serialized)

[maazbin]

Thanks for the code. It worked and solved my problem.