Extract GPSMAP data

main.py

import cv2
import numpy as np
import pafy
import pytesseract


TESSERACT_PARAMETERS = '--oem 3 --psm 6 -c tessedit_char_whitelist="Ckmtv0123456789.:"'


def cleanup_dist(dist: str) -> str:
    separator_position = dist.find(".")
    if separator_position != -1:
        return dist[:separator_position+3]
    return "?"


def cleanup_temperature(temperature: str) -> str:
    temperature = temperature.replace(",", ".")
    separator_position = temperature.find(".")
    if separator_position != -1:
        return f"{temperature[:separator_position+3]} °C"
    return "?"


def cleanup_speed(speed: str) -> str:
    speed = speed.replace("_", "").replace(" ", "").replace(",", ".")
    separator_position = speed.find(".")
    if separator_position != -1:
        return f"{speed[:separator_position+3]} kt"
    return "?"


def cleanup_v(v: str) -> str:
    v = v.replace(",", ".").replace(":", ".")
    separator_position = v.find(".")
    if separator_position != -1:
        return f"{v[:separator_position+2]} v"
    return "?"


def main():
    # TODO: Update video URL with your own video
    url = ""
    video = pafy.new(url)
    videoplay = video.getbest(preftype="any")
    cap = cv2.VideoCapture(videoplay.url)
    print(f"Video FPS: {cap.get(cv2.CAP_PROP_FPS)}")
    # In seconds
    start_time = 8 * 60
    end_time = 9 * 60
    # Set the start time
    cap.set(cv2.CAP_PROP_POS_MSEC, start_time * 1000)
    # Execute specific duration
    while True and cap.get(cv2.CAP_PROP_POS_MSEC) <= end_time * 1000:
        success, img = cap.read()
        if not success:
            break
        # Pre-process image
        img = img[0:360, 0:360]
        gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
        gray = cv2.bitwise_not(gray)
        blur = cv2.GaussianBlur(gray, (5, 5), 0)
        _, gray = cv2.threshold(blur, 40, 255, cv2.THRESH_BINARY)
        sharpen_kernel = np.array([[-1, -1, -1], [-1, 9, -1], [-1, -1, -1]])
        gray = cv2.filter2D(gray, -1, sharpen_kernel)

        # Show pre-processed image to screen
        cv2.imshow("Image", gray)
        cv2.waitKey(1)

        # Crop first image
        cropped_img = gray[0:113, 0:360]
        # Scale to 25%
        width = int(cropped_img.shape[1] * 25 / 100)
        height = int(cropped_img.shape[0] * 25 / 100)
        resized_img = cv2.resize(cropped_img, (width, height), interpolation = cv2.INTER_AREA)
        data1 = pytesseract.image_to_string(resized_img, lang='eng', config=TESSERACT_PARAMETERS).lower()
        data1 = data1.split('\n')
        # Cleanup data
        while "" in data1:
            data1.remove("")
        if len(data1) == 1 and "m" in data1[0]:
            data1[0] = cleanup_dist((data1[0]))
        else:
            data1 = ["?"]

        # Crop second image
        cropped_img = gray[113:360, 0:360]
        # Extract text with tesseract
        data2 = pytesseract.image_to_string(cropped_img, lang='eng', config=TESSERACT_PARAMETERS).lower()
        data2 = data2.split('\n')
        # Cleanup data
        while "" in data2:
            data2.remove("")
        # Skip broken data
        if len(data2) != 4:
            continue
        if ":" not in data2[3]:
            continue
        data2[0] = cleanup_temperature(data2[0])
        data2[1] = cleanup_speed(data2[1])
        data2[2] = cleanup_v(data2[2])
        print(data1 + data2)


if __name__ == '__main__':
    main()

1. Install youtube-dl via pip install youtube_dl

2. Install pafy via pip install git+https://github.com/mps-youtube/pafy.git. This is a workaround for mps-youtube/pafy#306.

3. Install opencv-python via pip install opencv-python

4. In MacOS, tesseract can be installed via brew install tesseract

5. Install pytesseract via pip install pytesseract