How to deploy a transformer model with Gradio

Question

I want to deploy a Huggingface tranformer object detection model with Gradio. I have followed exactly the Gradio documentation for transformer models, but I keep getting the error message below every time I try to launch the app:

ValueError                                Traceback (most recent call last)
<ipython-input-21-ae95d9ecbd73> in <module>()
----> 1 gr.Interface.from_pipeline(obj_detect).launch(share = True)

1 frames
/usr/local/lib/python3.7/dist-packages/gradio/external.py in load_from_pipeline(pipeline)
    575         }
    576     else:
--> 577         raise ValueError("Unsupported pipeline type: {}".format(type(pipeline)))
    578 
    579     # define the function that will be called by the Interface

ValueError: Unsupported pipeline type: <class 'transformers.pipelines.object_detection.ObjectDetectionPipeline'>

This is my code:

Import timm
Import gradio as gr
from transformers import pipeline
obj_detect = pipeline("object-detection")
gr.Interface.from_pipeline(obj_detect).launch(share = True)

And when I actually define a function, and launch the app, the app launches but returns an error when I try to make a prediction. Here is my function wrapper:

def Object_detector(image):
  obj_detect = pipeline("object-detection")
  detect = obj_detect(image)
  return detect

app = gr.Interface(Object_detector,inputs = "image",outputs = "label")
app.launch(share = True)

This is strange to me, because I have followed the documentation exactly, yet I am getting an error. Please, your advice will be very much appreciated.

How to deploy a transformer model with Gradio

0 Answers0