app.py

from sentence_transformers import SentenceTransformer
from sklearn.metrics.pairwise import cosine_similarity
import numpy as np
import matplotlib.pyplot as plt
import gradio as gr

model = SentenceTransformer('clip-ViT-L-14')

def predict(im1, im2):
  sim = cosine_similarity(model.encode(im1).reshape(1,-1), model.encode(im2).reshape(1,-1))
  if sim > 0.8:
    return sim, "SAME PERSON, UNLOCK PHONE"
  else:
    return sim, "DIFFERENT PEOPLE, DON'T UNLOCK"
    

interface = gr.Interface(fn=predict, 
                         inputs= [gr.Image(type="pil", source="webcam"), 
                                  gr.Image(type="pil", source="webcam")], 
                         outputs= [gr.Number(label="Similarity"),
                                   gr.Textbox(label="Message")]
                         )

interface.launch(debug=True)