import streamlit as st
from process_video import segment_video, classify
from io import StringIO
import cv2 as cv
import tempfile
import numpy as np
import tensorflow as tf
from crop_hand_skeleton import crop_hand
from cvzone.HandTrackingModule import HandDetector

if __name__ == "__main__":
	detector = HandDetector(maxHands=1, mode=True, detectionCon=0.7, minTrackCon=0.8)
	model = tf.keras.models.load_model('model_pred/VGG16_sign_char_detection_model')
	
	st.set_page_config(
		page_title="Projekt widzenie"
	)
	st.title("Projekt rozpoznawanie liter z alfabetu znaków migowych z wideo")

	st.write('Załaduj film')
    
	upload_movie = st.file_uploader("Wybierz film", type=["mp4"])
    
	if upload_movie:
		st.write("Film się ładuje.....")
		tfile = tempfile.NamedTemporaryFile(delete=False)
		tfile.write(upload_movie.read())
		video_cap = cv.VideoCapture(tfile.name)
		result, num = segment_video(video_cap, fps=10)
		st.write(f"Załadowano {num} klatek")
		for img in result:
			img_skeleton = crop_hand(img, detector)
			img2= cv.resize(img_skeleton,dsize=(224,224))
			#breakpoint()
			img_np = np.asarray(img2)
			classification = classify(img_np[:,:,::-1], model)
			st.image(img_skeleton[:,:,::-1])
			st.write(classification)