Automating Douyin Likes with Python, ADB, and Baidu AI Face Recognition

This guide explains how to create a Douyin (TikTok) automation script that automatically likes videos featuring attractive female faces. The solution combines Python, Android Debug Bridge (ADB) commands, and Baidu AI's face‑recognition API.

Prerequisites : Python environment (Anaconda recommended), ADB installed and device debugging enabled, and a Baidu AI account with appid , api_key , and secret_key for the face‑recognition service.

ADB basics : ADB acts as a bridge to control Android apps. Key commands used are:

# Launch Douyin app
adb shell am start -n com.ss.android.ugc.aweme com.ss.android.ugc.aweme.splash.SplashActivity

# Simulate a tap on the like button (example coordinates)
adb shell input tap 1330 1750

# Swipe up to the next video
adb shell input swipe 900 1400 400 1400 100

The script defines the package and activity names:

# Douyin app package and initial activity
package_name = 'com.ss.android.ugc.aweme'
activity_name = 'com.ss.android.ugc.aweme.splash.SplashActivity'

Python ADB control functions handle launching the app, taking screenshots, saving liked videos, and swiping to the next video:

import os
from PIL import Image

def start_my_app(package_name, activity_name):
    os.popen('adb shell am start -n %s/%s' % (package_name, activity_name))

def save_video_met(screen_name, find_girl_num):
    img = Image.open(screen_name).convert('RGB')
    img.save("漂亮的小姐姐/DYGirl_%d.jpg" % find_girl_num)
    os.system('adb shell input tap 1330 1750')

def play_next_video():
    os.system('adb shell input swipe 540 1300 540 500 100')

def get_screen_shot_part_img(image_name):
    os.system('adb shell /system/bin/screencap -p /sdcard/screenshot.jpg')
    os.system('adb pull /sdcard/screenshot.jpg %s' % image_name)
    img = Image.open(image_name).convert('RGB')
    w, h = img.size
    img = img.crop((0, 400, 1200, 2750))
    img.thumbnail((int(w/1.5), int(h/1.5)))
    img.save(image_name)
    return image_name

Main crawling logic repeatedly captures screenshots, sends them to Baidu AI for face analysis, and decides whether to like the video based on gender and a beauty score threshold (≥70). If a suitable face is found, the script likes the video and moves on; otherwise it continues scanning until a timeout is reached.

import datetime, time, shutil

if __name__ == '__main__':
    access_token = get_access_token()
    RECOGNITE_TOTAL_TIME = 10
    TYPE_IMAGE_LOCAL = 1
    start_my_app(package_name, activity_name)
    time.sleep(5)
    find_girl_num = 0
    while True:
        recognite_time_start = datetime.datetime.now()
        recognite_count = 1
        while True:
            screen_name = get_screen_shot_part_img('images/temp%d.jpg' % recognite_count)
            recognite_result = analysis_face(parse_face_pic(screen_name, TYPE_IMAGE_LOCAL, access_token))
            recognite_count += 1
            recognite_time_end = datetime.datetime.now()
            if recognite_result:
                find_girl_num += 1
                save_video_met(screen_name, find_girl_num)
                print('已经发现 %d 个漂亮小姐姐' % find_girl_num)
                break
            else:
                if (recognite_time_end - recognite_time_start).seconds < RECOGNITE_TOTAL_TIME:
                    continue
                else:
                    print('超时！！！这是一条没有吸引力的视频！')
                    break
        shutil.rmtree('./images')
        time.sleep(0.05)
        os.mkdir('./images')
        print('==' * 30)
        time.sleep(2)
        print('准备播放下一个视频~')
        play_next_video()
        time.sleep(2)

Adjust the screen coordinates in the ADB commands to match your device's resolution. The author notes that after a few days of running the crawler, Douyin's recommendation algorithm starts showing predominantly attractive videos, confirming the effectiveness of the automated liking strategy.

Conclusion : By integrating ADB automation with AI‑driven face analysis, developers can build a self‑sustaining Douyin bot that selectively engages with content, showcasing a practical application of mobile automation and computer vision.