نحوه استفاده از Cloud Run Jobs & APIهای هوش ویدیویی برای پردازش ویدیوها

۱. مقدمه

نمای کلی

در این آزمایشگاه کد، شما یک پروژه Cloud Run نوشته شده با Node.js ایجاد خواهید کرد که شرح بصری از هر صحنه در یک ویدیو ارائه می‌دهد. ابتدا، پروژه شما از API هوش ویدیویی برای تشخیص مهرهای زمانی برای هر تغییر صحنه استفاده خواهد کرد. در مرحله بعد، پروژه شما از یک فایل باینری شخص ثالث به نام ffmpeg برای گرفتن اسکرین شات برای هر مهر زمانی تغییر صحنه استفاده خواهد کرد. در نهایت، از کپشن نویسی بصری Vertex AI برای ارائه شرح بصری از اسکرین شات‌ها استفاده می‌شود.

این آزمایشگاه کد همچنین نحوه استفاده از ffmpeg را در Cloud Run Job شما برای ضبط تصاویر از یک ویدیو در یک زمان مشخص نشان می‌دهد. از آنجایی که ffmpeg باید به طور مستقل نصب شود، این آزمایشگاه کد به شما نشان می‌دهد که چگونه یک Dockerfile برای نصب ffmpeg به عنوان بخشی از Cloud Run Job خود ایجاد کنید.

در اینجا تصویری از نحوه عملکرد Cloud Run Job ارائه شده است:

تصویر شرح ویدیوی شغل Cloud Run

آنچه یاد خواهید گرفت

  • نحوه ایجاد یک تصویر کانتینر با استفاده از Dockerfile برای نصب یک فایل باینری شخص ثالث
  • چگونه با ایجاد یک حساب کاربری سرویس برای Cloud Run Job، اصل حداقل امتیاز را رعایت کنیم تا سایر سرویس‌های Google Cloud را فراخوانی کنیم؟
  • نحوه استفاده از کتابخانه کلاینت Video Intelligence از یک پروژه Cloud Run
  • نحوه فراخوانی API های گوگل برای دریافت توضیحات بصری هر صحنه از Vertex AI

۲. تنظیمات و الزامات

پیش‌نیازها

فعال کردن پوسته ابری

  1. از کنسول ابری، روی فعال کردن پوسته ابری کلیک کنید d1264ca30785e435.png .

cb81e7c8e34bc8d.png

اگر این اولین باری است که Cloud Shell را اجرا می‌کنید، یک صفحه میانی برای توضیح آن به شما نمایش داده می‌شود. اگر با یک صفحه میانی مواجه شدید، روی ادامه کلیک کنید.

d95252b003979716.png

آماده‌سازی و اتصال به Cloud Shell فقط چند لحظه طول می‌کشد.

7833d5e1c5d18f54.png

این ماشین مجازی مجهز به تمام ابزارهای توسعه مورد نیاز است. این ماشین یک دایرکتوری خانگی پایدار ۵ گیگابایتی ارائه می‌دهد و در فضای ابری گوگل اجرا می‌شود که عملکرد شبکه و احراز هویت را تا حد زیادی افزایش می‌دهد. بخش عمده‌ای از کار شما در این آزمایشگاه کد، اگر نگوییم همه، را می‌توان با یک مرورگر انجام داد.

پس از اتصال به Cloud Shell، باید ببینید که احراز هویت شده‌اید و پروژه روی شناسه پروژه شما تنظیم شده است.

  1. برای تأیید احراز هویت، دستور زیر را در Cloud Shell اجرا کنید:
gcloud auth list

خروجی دستور

 Credentialed Accounts
ACTIVE  ACCOUNT
*       <my_account>@<my_domain.com>

To set the active account, run:
    $ gcloud config set account `ACCOUNT`
  1. دستور زیر را در Cloud Shell اجرا کنید تا تأیید کنید که دستور gcloud از پروژه شما اطلاع دارد:
gcloud config list project

خروجی دستور

[core]
project = <PROJECT_ID>

اگر اینطور نیست، می‌توانید با این دستور آن را تنظیم کنید:

gcloud config set project <PROJECT_ID>

خروجی دستور

Updated property [core/project].

۳. فعال کردن APIها و تنظیم متغیرهای محیطی

قبل از اینکه بتوانید از این codelab استفاده کنید، باید چندین API را فعال کنید. این codelab به استفاده از APIهای زیر نیاز دارد. می‌توانید با اجرای دستور زیر این APIها را فعال کنید:

gcloud services enable run.googleapis.com \
    storage.googleapis.com \
    cloudbuild.googleapis.com \
    videointelligence.googleapis.com \
    aiplatform.googleapis.com

سپس می‌توانید متغیرهای محیطی را که در سراسر این آزمایشگاه کد استفاده خواهند شد، تنظیم کنید.

REGION=<YOUR-REGION>
PROJECT_ID=<YOUR-PROJECT-ID>
PROJECT_NUMBER=$(gcloud projects describe $PROJECT_ID --format='value(projectNumber)')
JOB_NAME=video-describer-job
BUCKET_ID=$PROJECT_ID-video-describer
SERVICE_ACCOUNT="cloud-run-job-video"
SERVICE_ACCOUNT_ADDRESS=$SERVICE_ACCOUNT@$PROJECT_ID.iam.gserviceaccount.com

۴. یک حساب کاربری سرویس ایجاد کنید

شما یک حساب کاربری سرویس برای Cloud Run Job ایجاد خواهید کرد تا از آن برای دسترسی به Cloud Storage، Vertex AI و Video Intelligence API استفاده کنید.

ابتدا، حساب کاربری سرویس را ایجاد کنید.

gcloud iam service-accounts create $SERVICE_ACCOUNT \
  --display-name="Cloud Run Video Scene Image Describer service account"

سپس به حساب کاربری سرویس، دسترسی به مخزن ذخیره‌سازی ابری و APIهای هوش مصنوعی ورتکس را اعطا کنید.

# to view & download storage bucket objects
gcloud projects add-iam-policy-binding $PROJECT_ID \
  --member serviceAccount:$SERVICE_ACCOUNT_ADDRESS \
  --role=roles/storage.objectViewer

# to call the Vertex AI imagetext model
gcloud projects add-iam-policy-binding $PROJECT_ID \
  --member serviceAccount:$SERVICE_ACCOUNT_ADDRESS \
  --role=roles/aiplatform.user

۵. یک فضای ذخیره‌سازی ابری ایجاد کنید

با دستور زیر یک فضای ذخیره‌سازی ابری ایجاد کنید که در آن بتوانید ویدیوها را برای پردازش توسط Cloud Run Job آپلود کنید:

gsutil mb -l us-central1 gs://$BUCKET_ID/

[اختیاری] می‌توانید با دانلود این ویدیوی نمونه به صورت محلی، از آن استفاده کنید.

gsutil cp gs://cloud-samples-data/video/visionapi.mp4 testvideo.mp4

حالا فایل ویدیویی خود را در فضای ذخیره‌سازی خود آپلود کنید.

FILENAME=<YOUR-VIDEO-FILENAME>
gsutil cp $FILENAME gs://$BUCKET_ID

۶. ایجاد شغل Cloud Run

ابتدا، یک دایرکتوری برای کد منبع ایجاد کنید و با دستور cd به آن دایرکتوری بروید.

mkdir video-describer-job && cd $_

سپس، یک فایل package.json با محتوای زیر ایجاد کنید:

{
  "name": "video-describer-job",
  "version": "1.0.0",
  "private": true,
  "description": "describes the image in every scene for a given video",
  "main": "app.js",
  "author": "Google LLC",
  "license": "Apache-2.0",
  "scripts": {
    "start": "node app.js"
  },
  "dependencies": {
    "@google-cloud/storage": "^7.7.0",
    "@google-cloud/video-intelligence": "^5.0.1",
    "axios": "^1.6.2",
    "fluent-ffmpeg": "^2.1.2",
    "google-auth-library": "^9.4.1"
  }
}

این برنامه برای بهبود خوانایی، از چندین فایل منبع تشکیل شده است. ابتدا، یک فایل منبع app.js با محتوای زیر ایجاد کنید. این فایل شامل نقطه ورود برای کار و منطق اصلی برنامه است.

const bucketName = "<YOUR_BUCKET_ID>";
const videoFilename = "<YOUR-VIDEO-FILENAME>";

const { captureImages } = require("./helpers/imageCapture.js");
const { detectSceneChanges } = require("./helpers/sceneDetector.js");
const { getImageCaption } = require("./helpers/imageCaptioning.js");
const storageHelper = require("./helpers/storage.js");
const authHelper = require("./helpers/auth.js");

const fs = require("fs").promises;
const path = require("path");

const main = async () => {

    try {

        // download the file to locally to the Cloud Run Job instance
        let localFilename = await storageHelper.downloadVideoFile(
            bucketName,
            videoFilename
        );

        // PART 1 - Use Video Intelligence API
        // detect all the scenes in the video & save timestamps to an array

        // EXAMPLE OUTPUT
        // Detected scene changes at the following timestamps:
        // [1, 7, 11, 12]
        let timestamps = await detectSceneChanges(localFilename);
        console.log(
            "Detected scene changes at the following timestamps: ",
            timestamps
        );

        // PART 2 - Use ffmpeg via dockerfile install
        // create an image of each scene change
        // and save to a local directory called "output"
        // returns the base filename for the generated images

        // EXAMPLE OUTPUT
        // creating screenshot for scene:  1 at output/video-filename-1.png
        // creating screenshot for scene:  7 at output/video-filename-7.png
        // creating screenshot for scene:  11 at output/video-filename-11.png
        // creating screenshot for scene:  12 at output/video-filename-12.png
        // returns the base filename for the generated images
        let imageBaseName = await captureImages(localFilename, timestamps);

        // PART 3a - get Access Token to call Vertex AI APIs via REST
        // needed for the image captioning
        // since we're calling the Vertex AI APIs directly
        let accessToken = await authHelper.getAccessToken();
        console.log("got an access token");

        // PART 3b - use Image Captioning to describe each scene per screenshot
        // EXAMPLE OUTPUT
        /*
        [
            {
                timestamp: 1,
                description:
                    "an aerial view of a city with a bridge in the background"
            },
            {
                timestamp: 7,
                description:
                    "a man in a blue shirt sits in front of shelves of donuts"
            },
            {
                timestamp: 11,
                description:
                    "a black and white photo of people working in a bakery"
            },
            {
                timestamp: 12,
                description:
                    "a black and white photo of a man and woman working in a bakery"
            }
        ]; */

        // instantiate the data structure for storing the scene description and timestamp
        // e.g. an array of json objects,
        // [{ timestamp: 5, description: "..." }, ...]
        let scenes = [];

        // for each timestamp, send the image to Vertex AI
        console.log("getting Vertex AI description for each timestamps");
        scenes = await Promise.all(
            timestamps.map(async (timestamp) => {
                let filepath = path.join(
                    "./output",
                    imageBaseName + "-" + timestamp + ".png"
                );

                // get the base64 encoded image bc sending via REST
                const encodedFile = await fs.readFile(filepath, "base64");

                // send each screenshot to Vertex AI for description
                let description = await getImageCaption(
                    accessToken,
                    encodedFile
                );

                return { timestamp: timestamp, description: description };
            })
        );

        console.log("finished collecting all the scenes");
        console.log(scenes);
    } catch (error) {
        //return an error
        console.error("received error: ", error);
    }
};

// Start script
main().catch((err) => {
    console.error(err);
});

در مرحله بعد، Dockerfile ایجاد کنید.

# Copyright 2020 Google, LLC.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#    http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

# Use the official lightweight Node.js image.
# https://hub.docker.com/_/node
FROM node:20.10.0-slim

# Create and change to the app directory.
WORKDIR /usr/src/app

RUN apt-get update && apt-get install -y ffmpeg

# Copy application dependency manifests to the container image.
# A wildcard is used to ensure both package.json AND package-lock.json are copied.
# Copying this separately prevents re-running npm install on every code change.
COPY package*.json ./

# Install dependencies.
# If you add a package-lock.json speed your build by switching to 'npm ci'.
# RUN npm ci --only=production
RUN npm install --production

# Copy local code to the container image.
COPY . .

# Run the job on container startup.
CMD [ "npm", "start" ]

و فایلی به نام .dockerignore ایجاد کنید تا از کانتینری کردن فایل‌های خاص صرف نظر شود.

Dockerfile
.dockerignore
node_modules
npm-debug.log

حالا یک پوشه به نام helpers ایجاد کنید. این پوشه شامل ۵ فایل کمکی خواهد بود.

mkdir helpers
cd helpers

در مرحله بعد، یک فایل sceneDetector.js با محتوای زیر ایجاد کنید. این فایل از API هوش ویدیویی برای تشخیص تغییر صحنه‌ها در ویدیو استفاده می‌کند.

const fs = require("fs");
const util = require("util");
const readFile = util.promisify(fs.readFile);
const ffmpeg = require("fluent-ffmpeg");

const Video = require("@google-cloud/video-intelligence");
const client = new Video.VideoIntelligenceServiceClient();

module.exports = {
    detectSceneChanges: async function (downloadedFile) {
        // Reads a local video file and converts it to base64
        const file = await readFile(downloadedFile);
        const inputContent = file.toString("base64");

        // setup request for shot change detection
        const videoContext = {
            speechTranscriptionConfig: {
                languageCode: "en-US",
                enableAutomaticPunctuation: true
            }
        };

        const request = {
            inputContent: inputContent,
            features: ["SHOT_CHANGE_DETECTION"]
        };

        // Detects camera shot changes
        const [operation] = await client.annotateVideo(request);
        console.log("Shot (scene) detection in progress...");
        const [operationResult] = await operation.promise();

        // Gets shot changes
        const shotChanges =
            operationResult.annotationResults[0].shotAnnotations;

        console.log(
            "Shot (scene) changes detected: " + shotChanges.length
        );

        // data structure to be returned
        let sceneChanges = [];

        // for the initial scene
        sceneChanges.push(1);

        // if only one scene, keep at 1 second
        if (shotChanges.length === 1) {
            return sceneChanges;
        }

        // get length of video
        const videoLength = await getVideoLength(downloadedFile);

        shotChanges.forEach((shot, shotIndex) => {
            if (shot.endTimeOffset === undefined) {
                shot.endTimeOffset = {};
            }
            if (shot.endTimeOffset.seconds === undefined) {
                shot.endTimeOffset.seconds = 0;
            }
            if (shot.endTimeOffset.nanos === undefined) {
                shot.endTimeOffset.nanos = 0;
            }

            // convert to a number
            let currentTimestampSecond = Number(
                shot.endTimeOffset.seconds
            );

            let sceneChangeTime = 0;
            // double-check no scenes were detected within the last second
            if (currentTimestampSecond + 1 > videoLength) {
                sceneChangeTime = currentTimestampSecond;
            } else {
                // otherwise, for simplicity, just round up to the next second
                sceneChangeTime = currentTimestampSecond + 1;
            }

            sceneChanges.push(sceneChangeTime);
        });

        return sceneChanges;
    }
};

async function getVideoLength(localFile) {
    let getLength = util.promisify(ffmpeg.ffprobe);
    let length = await getLength(localFile);

    console.log("video length: ", length.format.duration);
    return length.format.duration;
}

حالا فایلی به نام imageCapture.js با محتوای زیر ایجاد کنید. این فایل از پکیج node fluent-ffmpeg برای اجرای دستورات ffmpeg از درون یک برنامه node استفاده می‌کند.

const ffmpeg = require("fluent-ffmpeg");
const path = require("path");
const util = require("util");

module.exports = {
    captureImages: async function (localFile, scenes) {
        let imageBaseName = path.parse(localFile).name;

        try {
            for (scene of scenes) {
                console.log("creating screenshot for scene: ", +scene);
                await createScreenshot(localFile, imageBaseName, scene);
            }
        } catch (error) {
            console.log("error gathering screenshots: ", error);
        }

        console.log("finished gathering the screenshots");
        return imageBaseName; // return the base filename for each image
    }
};

async function createScreenshot(localFile, imageBaseName, scene) {
    return new Promise((resolve, reject) => {
        ffmpeg(localFile)
            .screenshots({
                timestamps: [scene],
                filename: `${imageBaseName}-${scene}.png`,
                folder: "output",
                size: "320x240"
            })
            .on("error", () => {
                console.log(
                    "Failed to create scene for timestamp: " + scene
                );
                return reject(
                    "Failed to create scene for timestamp: " + scene
                );
            })
            .on("end", () => {
                return resolve();
            });
    });
}

در نهایت، فایلی به نام imageCaptioning.js با محتوای زیر ایجاد کنید. این فایل از Vertex AI برای دریافت توضیحات بصری از هر تصویر صحنه استفاده می‌کند.

const axios = require("axios");
const { GoogleAuth } = require("google-auth-library");

const auth = new GoogleAuth({
    scopes: "https://www.googleapis.com/auth/cloud-platform"
});

module.exports = {
    getImageCaption: async function (token, encodedFile) {
        // this example shows you how to call the Vertex REST APIs directly
        // https://cloud.google.com/vertex-ai/generative-ai/docs/image/image-captioning#get-captions-short
        // https://cloud.google.com/vertex-ai/generative-ai/docs/model-reference/image-captioning

        let projectId = await auth.getProjectId();

        let config = {
            headers: {
                "Authorization": "Bearer " + token,
                "Content-Type": "application/json; charset=utf-8"
            }
        };

        const json = {
            "instances": [
                {
                    "image": {
                        "bytesBase64Encoded": encodedFile
                    }
                }
            ],
            "parameters": {
                "sampleCount": 1,
                "language": "en"
            }
        };

        let response = await axios.post(
            "https://us-central1-aiplatform.googleapis.com/v1/projects/" +
                projectId +
                "/locations/us-central1/publishers/google/models/imagetext:predict",
            json,
            config
        );

        return response.data.predictions[0];
    }
};

فایلی به نام auth.js ایجاد کنید. این فایل از کتابخانه کلاینت احراز هویت گوگل برای دریافت توکن دسترسی مورد نیاز برای تماس مستقیم با نقاط انتهایی Vertex AI استفاده خواهد کرد.

const { GoogleAuth } = require("google-auth-library");

const auth = new GoogleAuth({
    scopes: "https://www.googleapis.com/auth/cloud-platform"
});

module.exports = {
    getAccessToken: async function () {
        return await auth.getAccessToken();
    }
};

در نهایت، فایلی به نام storage.js ایجاد کنید. این فایل از کتابخانه‌های کلاینت Cloud Storage برای دانلود ویدیو از فضای ابری استفاده خواهد کرد.

const { Storage } = require("@google-cloud/storage");

module.exports = {
    downloadVideoFile: async function (bucketName, videoFilename) {
        // Creates a client
        const storage = new Storage();

        // keep same name locally
        let localFilename = videoFilename;

        const options = {
            destination: localFilename
        };

        // Download the file
        await storage
            .bucket(bucketName)
            .file(videoFilename)
            .download(options);

        console.log(
            `gs://${bucketName}/${videoFilename} downloaded locally to ${localFilename}.`
        );

        return localFilename;
    }
};

۷. استقرار و اجرای کار Cloud Run

ابتدا مطمئن شوید که در دایرکتوری ریشه video-describer-job مربوط به codelab خود هستید.

cd .. && pwd

سپس، می‌توانید از این دستور برای استقرار Cloud Run Job استفاده کنید.

gcloud run jobs deploy $JOB_NAME  --source . --region $REGION

اکنون می‌توانید با اجرای دستور زیر، Cloud Run Job را اجرا کنید:

gcloud run jobs execute $JOB_NAME

پس از اتمام اجرای کار، می‌توانید دستور زیر را اجرا کنید تا لینکی به آدرس URL مربوط به لاگ دریافت کنید. (یا می‌توانید از کنسول ابری استفاده کنید و مستقیماً به Cloud Run Jobs بروید تا لاگ‌ها را ببینید.)

gcloud run jobs executions describe <JOB_EXECUTION_ID>

باید خروجی زیر را در لاگ‌ها مشاهده کنید:

[{ timestamp: 1, description: 'what is google cloud vision api ? is written on a white background .'},
{ timestamp: 3, description: 'a woman wearing a google cloud vision api shirt sits at a table'},
{ timestamp: 18, description: 'a person holding a cell phone with the words what is cloud vision api on the bottom' }, ...]

۸. تبریک می‌گویم!

تبریک می‌گویم که آزمایشگاه کد را تمام کردید!

توصیه می‌کنیم مستندات مربوط به Video Intelligence API ، Cloud Run و Vertex AI visual captioning را بررسی کنید.

آنچه ما پوشش داده‌ایم

  • نحوه ایجاد یک تصویر کانتینر با استفاده از Dockerfile برای نصب یک فایل باینری شخص ثالث
  • چگونه با ایجاد یک حساب کاربری سرویس برای Cloud Run Job، اصل حداقل امتیاز را رعایت کنیم تا سایر سرویس‌های Google Cloud را فراخوانی کنیم؟
  • نحوه استفاده از کتابخانه کلاینت Video Intelligence از یک پروژه Cloud Run
  • نحوه فراخوانی API های گوگل برای دریافت توضیحات بصری هر صحنه از Vertex AI

۹. تمیز کردن

برای جلوگیری از هزینه‌های ناخواسته، (برای مثال، اگر این کار Cloud Run سهواً بیشتر از تخصیص فراخوانی ماهانه Cloud Run شما در سطح رایگان فراخوانی شود)، می‌توانید کار Cloud Run یا پروژه‌ای را که در مرحله 2 ایجاد کرده‌اید، حذف کنید.

برای حذف کار Cloud Run، به کنسول Cloud Run در آدرس https://console.cloud.google.com/run/ بروید و تابع video-describer-job (یا $JOB_NAME را در صورتی که از نام دیگری استفاده کرده‌اید) حذف کنید.

اگر تصمیم به حذف کل پروژه دارید، می‌توانید به آدرس https://console.cloud.google.com/cloud-resource-manager بروید، پروژه‌ای را که در مرحله ۲ ایجاد کرده‌اید انتخاب کنید و گزینه Delete را انتخاب کنید. اگر پروژه را حذف کنید، باید پروژه‌ها را در Cloud SDK خود تغییر دهید. می‌توانید با اجرای gcloud projects list لیست تمام پروژه‌های موجود را مشاهده کنید.