app.py

import gradio as gr
import torch
from tensorflow.keras.utils import img_to_array
import imutils
import cv2
from keras.models import load_model
import numpy as np

# parameters for loading data and images
detection_model_path = 'haarcascade_frontalface_default.xml'
emotion_model_path = '_mini_XCEPTION.102-0.66.hdf5'
#emotion_model_path = 'emotion_detection_model_state.pth'

# hyper-parameters for bounding boxes shape
# loading models
face_detection = cv2.CascadeClassifier(detection_model_path)
#emotion_classifier = torch.load(emotion_model_path,map_location=torch.device('cpu'))
emotion_classifier = load_model(emotion_model_path, compile=False)
EMOTIONS = ["angry", "disgusted", "scared", "happy", "sad", "surprised",
            "neutral"]


def predict(frame):

    frame = imutils.resize(frame, width=300)
    gray = cv2.cvtColor(frame, cv2.COLOR_RGB2GRAY)
    faces = face_detection.detectMultiScale(gray, scaleFactor=1.1,
                                            minNeighbors=5, minSize=(30, 30),
                                            flags=cv2.CASCADE_SCALE_IMAGE)

    frameClone = frame.copy()
    if len(faces) > 0:
        faces = sorted(faces, reverse=True,
                       key=lambda x: (x[2] - x[0]) * (x[3] - x[1]))[0]
        (fX, fY, fW, fH) = faces
        # Extract the ROI of the face from the grayscale image, resize it to a fixed 28x28 pixels, and then prepare
        # the ROI for classification via the CNN
        roi = gray[fY:fY + fH, fX:fX + fW]
        roi = cv2.resize(roi, (64, 64))
        roi = roi.astype("float") / 255.0
        roi = img_to_array(roi)
        roi = np.expand_dims(roi, axis=0)

        preds = emotion_classifier.predict(roi)[0]
        label = EMOTIONS[preds.argmax()]
    else:
        return frameClone, "No se ha detectado ninguna cara"

    probs = {}
    cv2.putText(frameClone, label, (fX, fY - 10),
                cv2.FONT_HERSHEY_DUPLEX, 1, (238, 164, 64), 1)
    cv2.rectangle(frameClone, (fX, fY), (fX + fW, fY + fH),
                  (238, 164, 64), 2)

    for (i, (emotion, prob)) in enumerate(zip(EMOTIONS, preds)):
        probs[emotion] = float(prob)

    return frameClone, probs


inp = gr.inputs.Image(source="webcam")
out = [
    gr.outputs.Image(label="Emoción detectada"),
    gr.outputs.Label(num_top_classes=2, label="Posibles emociones detectadas")
]
title = "Reconocimiento de emociones"
article = "Es posible que una mala iluminación impida la detección del rostro"
description = "Este espacio muestra una demo del proyecto para Saturdays AI  (https://saturdays.ai/2021/11/22/reconocimiento-de-emociones/)."
gr.Interface(predict, inp, out, capture_session=True, title=title, article=article , theme="dark-peach",
             description=description).launch(inbrowser=True)