Detcting commercials and controlling the TV

main.py

from tv import TV
import time
import cv2
import pytesseract
import numpy as np
from PIL import ImageGrab

device_id = "<YOUR TV ID>"
access_token = "<YOUR ACCESS TOKEN>"
# Area on the screen to look for commercials break text
bounding_box = (55, 115, 280, 310)
# Commercials related texts
commercials_text = ["חוזרים", "בעוד"]
# How much time to wait after commercials end - this will prevent false unmutes if text is not recognized momentarily
unmute_delay = 5
commercials_frames_needed = 3

tv = TV(access_token, device_id)
options = "-l heb --oem 3 --psm 12 tessedit_char_whitelist=01234567890"
last_commercials_seen = 0
commercials_seen = 0

while True:
    cap_scr = np.array(ImageGrab.grab(bounding_box))
    cap_scr = cv2.cvtColor(cap_scr, cv2.COLOR_RGB2BGR)
    # extracted_text = pytesseract.image_to_string(cap_scr, config=options)
    # print(extracted_text)
    # Perform bounding box detection using Tesseract's built-in capabilities
    d = pytesseract.image_to_data(cap_scr, config=options, output_type=pytesseract.Output.DICT)
    n_boxes = len(d['text'])
    for i in range(n_boxes):
        if int(d['conf'][i]) > 50:
            if d['text'][i] in commercials_text:
                (x, y, w, h) = (d['left'][i], d['top'][i], d['width'][i], d['height'][i])
                cap_scr = cv2.rectangle(cap_scr, (x, y), (x + w, y + h), (0, 255, 0), 2)
                print(f"Found: {d['text'][i]} with conf {d['conf'][i]}")
                last_commercials_seen = time.time()
                commercials_seen += 1
                if commercials_seen > commercials_frames_needed:
                    tv.mute()
    if time.time() - last_commercials_seen > unmute_delay and commercials_seen > 0:
        commercials_seen = 0
        tv.mute(False)
    cv2.imshow("IP Camera Stream", cap_scr)
    if cv2.waitKey(1) & 0xFF == ord('q'):
        break