I tried to combine Discord Bot and face recognition-for LT-

Introduction

Materials for presentation at LT

I tried to make the face recognition (OpenCV) that I have done so far into a bot

• Face recognition articles have not been compiled yet (; ・ ∀ ・)

The item description is as follows

• Why Discord?
• Module, directory structure, etc.
• Discord Bot

Why Discord?

** Somehow! ** **

Until now, I've already made it with Slack Bot.

But I wonder if LINE Bot is ...

That's right, ** Discord can make bots, right? ?? ?? ** **

I researched Discord Bot

Official (discord.py)

It looks fun so I made it ☆

How to make a Discord bot has been updated the other day. (It's a little rough ...)

Module, directory structure, etc.

module

What you need is as follows Install what you need!

pip install discord       #Used when you want to do various things with discord
pip install opencv-python #Required when using OpenCV functions (used with cv2 on python)
pip install os            #You can use functions that depend on os
pip install requests      #Used for HTTP communication
pip install shutil        #High-level manipulation of files and collections of files

Directory structure

The directory structure is as follows

-- dis_test
   -- cascade
      -- haarcascade_frontalface_alt.xml
   -- img
   -- dis.py

Face learning data is here I used.

Discord Bot

program

dis_bot.py

# coding:utf-8

import discord
import time
import requests
import shutil
import cv2
import os

# from scripts.photo_processing import PhotoProcessing

IMG_PATH = "./img/"
IMG_SIZE = (128, 128)

FACE_CASCADE_PATH = "./cascade/haarcascade_frontalface_alt.xml"
FACE_CASCADE = cv2.CascadeClassifier(FACE_CASCADE_PATH)
TOKEN = "TOKEN ID"
client = discord.Client()

now_time = time.time()
times = time.gmtime(now_time)

#bot startup processing
@client.event
async def on_ready():
    channel = client.get_channel(Channel ID)
    await channel.send("Hi! how are you?")

#When receiving a message
@client.event
async def on_message(message):
    #Ignore if sender is a bot
    if message.author.bot:
        return
    
    if message.content.startswith('trim') and len(message.attachments) != 0:
        channel_id = message.channel.id
        url = message.attachments[0].url
        file_name = url.rsplit("/", 1)[1]

        #Save image
        rst = requests.get(url, stream = True)

        open_file = open(IMG_PATH + file_name, "wb")
        shutil.copyfileobj(rst.raw, open_file)
        open_file.close()

        #Image scrutiny
        img_file = cv2.imread(IMG_PATH + file_name, cv2.IMREAD_COLOR)

        # processing
        gray_file = cv2.cvtColor(img_file, cv2.COLOR_BGR2GRAY)
        front_face_list = FACE_CASCADE.detectMultiScale(gray_file, minSize=(50, 50), minNeighbors=3)

        if len(front_face_list) > 0:
            # trimming
            for (x, y, width, height) in front_face_list:

                trim_file = img_file[y:y+height, x:x+width]
                trim_file = cv2.resize(trim_file, IMG_SIZE)

                cv2.imwrite("{0}trim_{1}".format(IMG_PATH, file_name), trim_file)

                await message.channel.send('', file=discord.File("{0}trim_{1}".format(IMG_PATH, file_name)))
        else:
            await message.channel.send("The face was not authenticated.")

        #Image deletion
        os.remove(IMG_PATH + file_name)
        os.remove("{0}trim_{1}".format(IMG_PATH, file_name))

client.run(TOKEN)

Output result Command line

It recognizes the face properly and trims it

Of course, multiple people can trim

It seems that the cropped image can be used as an icon! (I don't use it) That's all for Discord Bot x Face Recognition (˘ω˘)