Spaces:
Sleeping
Sleeping
File size: 1,197 Bytes
4a32a5c 4769d33 523af2e 4a32a5c 4769d33 4a32a5c |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 |
from fastai.vision.all import *
import gradio as gr
import pathlib, os
classes = ['rock', 'paper', 'scissors'] # c0, c1, c2
def classify_image(img, model='rock-paper-scissors-resnet34.pkl'):
if os.name == 'nt': # workaround for Windows
pathlib.PosixPath = pathlib.WindowsPath
if os.name == 'posix': # workaround for Linux
pathlib.WindowsPath = pathlib.PosixPath
learn = load_learner(model)
pred,idx,probs = learn.predict(img)
return dict(zip(classes, map(float, probs)))
models = ['rock-paper-scissors-squeezenet.pkl','rock-paper-scissors-resnet34.pkl']
model = gr.Dropdown(models, label="Select Model")
image = gr.inputs.Image(shape=(192,192))
label = gr.outputs.Label()
examples = [
['c0-rock-IMG_20230225_171937.jpg'],
['c0-rock-IMG_20230225_171940.jpg'],
['c1-paper-IMG_20230225_172010.jpg'],
['c1-paper-IMG_20230225_172018.jpg'],
['c2-scissors-IMG_20230225_172025.jpg'],
['c2-scissors-IMG_20230225_172033.jpg']
]
# iface = gr.Interface(fn=classify_image, inputs=image, outputs=label, examples=examples)
iface = gr.Interface(fn=classify_image, inputs=[image, model], outputs=label, examples=examples)
iface.launch() |