main.py

import os
from ulti import *
import threading
# from streamlit_webrtc import (
#     AudioProcessorBase,
#     ClientSettings,
#     VideoProcessorBase,
#     WebRtcMode,
#     webrtc_streamer,
# )
# import av

# WEBRTC_CLIENT_SETTINGS = ClientSettings(
#     rtc_configuration={"iceServers": [{"urls": ["stun:stun.l.google.com:19302"]}]},
#     media_stream_constraints={
#         "video": True,
#         "audio": False,
#     },)

model_path = os.path.join('model','ModelTrainOnKaggle.h5')
@st.cache
def model_load():
    model = tf.keras.models.load_model(model_path)
    return model

def main():
    st.set_page_config(layout="wide")
    
    st.image(os.path.join('Images','Banner No2.png'), use_column_width  = True)
    st.markdown("<h1 style='text-align: center; color: white;'>Time to become a comic book character</h1>", unsafe_allow_html=True)
    with st.beta_expander("Configuration Option"):

        st.write("**AutoCrop** help the model by finding and cropping the biggest face it can find.")
        st.write("**Gamma Adjustment** can be used to lighten/darken the image")
    comic_model = model_load()


    # menu = ['Image Based', 'Video Based']
    menu = ['Image Based']
    st.sidebar.header('Mode Selection')
    choice = st.sidebar.selectbox('How would you like to be turn ?', menu)

    # Create the Home page
    if choice == 'Image Based':
        
        st.sidebar.header('Configuration')
        outputsize = st.sidebar.selectbox('Output Size', [384,512,768])
        Autocrop = st.sidebar.checkbox('Auto Crop Image',value=True) 
        gamma = st.sidebar.slider('Gamma adjust', min_value=0.1, max_value=3.0,value=1.0,step=0.1) # change the value here to get different result
        
 

        Image = st.file_uploader('Upload your portrait here',type=['jpg','jpeg','png'])
        if Image is not None:
            col1, col2 = st.beta_columns(2)
            Image = Image.read()
            Image = tf.image.decode_image(Image, channels=3).numpy()                  
            Image = adjust_gamma(Image, gamma=gamma)
            with col1:
                st.image(Image)
            input_image = loadtest(Image,cropornot=Autocrop)
            prediction = comic_model(input_image, training=True)
            prediction = tf.squeeze(prediction,0)
            prediction = prediction* 0.5 + 0.5
            prediction = tf.image.resize(prediction, 
                           [outputsize, outputsize],
                           method=tf.image.ResizeMethod.NEAREST_NEIGHBOR)
            prediction=  prediction.numpy()
            with col2:
                st.image(prediction)
    

    elif choice == 'Video Based':

    #     class OpenCVVideoProcessor(VideoProcessorBase):
    #         def __init__(self) -> None:
    #             self._model_lock = threading.Lock()
    #             self.model = model_load()
            
    #         def recv(self, frame: av.VideoFrame):

    #             img = frame.to_ndarray(format="bgr24")
    #             img = cv2.flip(img, 1)
    #             frame =loadframe(img)
    #             frame = self.model(frame, training=True)
    #             frame = tf.squeeze(frame,0)
    #             frame = frame* 0.5 + 0.5
    #             frame = tf.image.resize(frame, 
    #                         [384, 384],
    #                         method=tf.image.ResizeMethod.NEAREST_NEIGHBOR)
    #             frame = frame.numpy()
    #             print(type(frame))
    #             print(frame.shape)

    #             return av.VideoFrame.from_ndarray(frame, format="bgr24")

        
    #     webrtc_streamer(key="Test",
    #     client_settings=WEBRTC_CLIENT_SETTINGS,
    #     async_processing=True,video_processor_factory=OpenCVVideoProcessor,

    # )
        run = st.checkbox('Run',Value=true)
        FRAMEWINDOW = st.image([])
        camera = cv2.VideoCapture(0)
        gamma = st.slider('Gamma adjust', min_value=0.1, max_value=3.0,value=1.0,step=0.1)
        while run:
            _ , frame = camera.read()
            frame = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
            frame  = cv2.flip(frame, 1)
            frame = adjust_gamma(frame, gamma=gamma)
            # Framecrop = st.checkbox('Auto Crop Frame')
            frame = loadframe(frame)
            frame = comic_model(frame, training=True)
            frame = tf.squeeze(frame,0)
            frame = frame* 0.5 + 0.5
            frame = tf.image.resize(frame, 
                            [384, 384],
                            method=tf.image.ResizeMethod.NEAREST_NEIGHBOR)
            frame = frame.numpy()
            FRAMEWINDOW.image(frame)

if __name__ == '__main__':
    main()