Node Frame Extractor

This simeple command line utility extracts the specified frames (and their adjecent frames) from a video, resizes them and uploads them to a bucket in Amazon S3.

Usage

extract-frames --input [video] --data [data.json] --fps [framerate] --push-to-cloud

For a detailed list of available options and defaults, run the following:

extract-frames --help

Example

The following example dumps six adjecent frames (three on either sides) of all given frames using a framerate of 29.97 and push them to the cloud storage.

extract-frames --input video.mp4 --data frames.json --fps 29.97 --frame-count 3 --push-to-cloud

Amazon S3 Credentials

This utility uses Amazon AWS credentials information stored in the current users ~/.aws/credentials file. If you have more than one credentials configuration in the file, you can use the --aws-profile command line switch to specify which profile to use.

For example:

extract-frames --input myVideo.mp4 --data frames.json --aws-profile myprofilename --push-to-cloud

For more information on how to setup local credentials, refer to Amazon's AWS documentation.

Parameters

This command line utility supports the following parameters.

| Switch | Default | Description | | ------ | ------- | ----------- | | --input, -i | required | Input video filename. | | --data, -d | required if not -a | Scene data in Json format with a scene object having startFrame and endFrame keys. | | --fps, -r | 23.976 | Framerate of the input video. Must be accurate to extract scenes properly. | | --frame-count, -c | 5 | Number of frames to extract on either side of the scene. This is in addition to the frame marked in the scene boundry. | | --concurrency, -t | 8 | Number of concurrent sub-processes to use simultaneously. Should be used to tweak resource consumption. Use a value of 0 to run all sub-processes at once. | | --bucket, -b | ingrain.scenes.frames | S3 Bucket to which the extracted frames will be pushed to. | | --push-to-cloud, -p | false | Determines whether the extracted images be pushed to a cloud storage or not. | | --extract-all-frames, -a | false | Extract all frames of the video. Requires --fps and --total-frames. | | --total-frames, -f | null | Total number of frames to extract from the video. | | --video-id, -v | null | An identifier for the video being processed. It is used as an identifier in directory name. | | --aws-profile | default | Name of the AWS Credentials profile to use from the ~/.aws/credentials file. | | --aws-region | us-west-2 | Specify the AWS region to use. |

Installation

Prerequisites

You should have ffmpeg and graphicsmagik installed on your machine, and they should be exposed in the environment PATH.

Via NPM

The simplest way to install node-frame-extractor is via NPM. All you need to do is:

npm install -g node-frame-extractor

After that you can use the extract-frames command anywhere in your system.

Via Git Repository

If you rather install from source, you can clone the repository and set it up manually.

Step 1: Clone the repository.

git clone https://github.com/UzEE/node-frame-extractor.git

Step 2: Install the project dependencies.

cd ingrain-frame-ext
npm install 

Step 3: Manually prepare the CLI utility.

grunt prepare

Step 4: Link the local module to your global NPM cache.

npm link

You should now be able to use extract-frames command from anywhere in your system.

Input Data Format

Extract Specific Frames

This utility expects the input --data to have certain specific keys:

| field | value | | ----- | ----- | | videoId | The numeric ID of the video being processed. It is used as a part of the name of the AWS container. | | frames | An array of all the frames to process. |

For example:

{
	"videoId": 7,
	"frames": [14, 53, 234, 539, 872, 1143, 2763, 3234]
}

Note: You can also use the --video-id or -v switch to pass in the Video ID.

Extract All Frames

If you want to dump all frames of the video instead, you only need to pass the following parameters instead of the --data option:

| field | value | | ----- | ----- | | -v, --video-id | The numeric ID of the video being processed. It is used as a part of the name for the directory. A random number will be used if omitted. |

Uploaded Asset URIs

The uploaded assets can be accessed from the Amazon S3 bucket using the following URI format:

<bucket uri>/video.<video id>/frame.<frame number>.jpg

For example, the URI for frame 4104 of video Id 1 in the ingrain.scenes.frames bucket would resolve to:

https://s3-us-west-2.amazonaws.com/ingrain.scenes.frames/video.1/frame.4104.jpg