app.py

import cv2
import numpy as np
import matplotlib.pyplot as plt
from PIL import Image
import tensorflow as tf  
import tf_keras
import gradio as gr
import requests
import os

model_url = "https://huggingface.co/foryahasake/act-h5/resolve/main/bestXceptionPlusData.h5"
model_filename = "bestXceptionPlusData.h5"
from huggingface_hub import from_pretrained_keras

model = from_pretrained_keras( "foryahasake/act-h5")

def inference(inp):
    if len(inp.shape) == 3:  # If the image is colored
        pil_gray = cv2.cvtColor(inp, cv2.COLOR_RGB2GRAY)
    else:
        pil_gray = inp
    resized_img = cv2.resize(pil_gray, (48,48), interpolation=cv2.INTER_CUBIC)

    model_input = resized_img / 255.0
    model_input = np.expand_dims(model_input, axis=-1)  # Add channel dimension
    model_input = np.expand_dims(model_input, axis=0)   # Add batch dimension

    # Prediction
    result = model.predict(model_input)
    predictions = result[0]
    predicted_class = np.argmax(predictions)
    emotion_labels = ["Angry", "Disgust", "Fear", "Happy", "Sad", "Surprise", "Neutral"]
    predicted_emotion = emotion_labels[predicted_class]
    confidence = predictions[predicted_class]
    return predicted_emotion

demo = gr.Interface(fn=inference, inputs="image", outputs="label")
demo.launch()