mediaorient is a CLI tool and Go library to calculate the orientation of
images & videos based on their content, not EXIF data.

You can use mediaorient in two ways: as a command-line interface (CLI) tool or a Go library.

The CLI tool is a standalone application that can be used to calculate and fix the orientation of media files, while the library can be integrated into your own Go projects.

The binaries are available for Windows, macOS, and Linux. Download the latest release that matches your computer architecture and operating system.

Run the command below in the terminal:

$ mediaorient files <media1> [<media2> ...]

Run the command below in the terminal:

$ mediaorient dir <directory> [-r] [--mt <media-type>]

Where:

• dir (mandatory): the path to the directory where the media files are located.
• -r (optional): recursively search for files in subdirectories to include in the calculation.
• --mt (optional): the file types to be included in the calculation. You can choose between image, video, or all (default).

Other parameters you can use:

• -o (optional): the output format; you can choose report (default) or, if you prefer a raw output, json or csv.
• --ie (optional): ignores errors and continues the calculation even if some files are not valid.

For the full list of parameters, type mediaorient --help in the terminal.

In its default configuration, the mediaorient library supports media files with the following extensions:

• Images: .bmp, .gif, .jpg (.jpeg), .png, .tiff, .webp
• Videos: .avi, .mp4 (.m4v), .mkv, .mov, .webm

The CLI supports two additional image formats: .avif and .heic.

If you want to work with additional file extensions in the library, like those two above, you can use the functions AddImageType or AddVideoType before performing any orientation calculation. This allows mediaorient to include these file types during calculations.

When adding support for new media formats, it's essential to load a 3rd party library capable of decoding them. For example, to enable AVIF image calculation in mediaorient, you could use a library like avif-go to do this:

import _ "github.com/vegidio/avif-go"
mediaorient.AddImageType(".avif")

This app uses a neural network trained with almost 1 million images, creating a model of nearly 100 MB. Unfortunately, it becomes impractical to embed this data in the executable, otherwise the file would become too big.

For this reason, the first time you run the app, it will download the model from the internet; this may take a few seconds, depending on your internet connection.

This approach not only prevents an excessive executable size but also allows the app to always use the most recent version of the model, which may contain improvements and bug fixes.

While the model has accuracy of about 99%, it isn’t perfect. It works very well with everyday media, such as photos of people, animals, documents and common scenes, but may perform less reliably with certain special cases.

For example, if you have a media depicting a medical procedure, the app may not detect the correct orientation because the model was not trained (or was only minimally trained), on this type of media. You can help improve the app by contributing to the training dataset. Upload your own media here, so that in future versions of the app this kind of media will be properly detected.

Any media you upload will not be shared with anyone else and will be used solely to train the neural network. I’m not here to judge anyone, so feel free to upload any legal content.

However, I cannot emphasize this enough: do NOT upload any illegal media. If you upload something illegal, I will not only block your access to this project, but I will also report you to the authorities using your IP address and any other information available.

If the orientation calculation of videos is not working, it may be because you don't have FFmpeg working in your computer, which is required to extract frames from the video files.

When FFmpeg is not found, mediaorient will try to automatically download and install it for you. Even though this will work in most cases, it may fail for unpredictable reasons.

The best option to have the video calculation working is to install FFmpeg yourself in your computer and make sure it is available in your PATH.

For a couple of years now, Microsoft and Apple have required developers to join their "Developer Program" to gain the pretentious status of an identified developer 😛.

Translating to non-BS language, this means that if you’re not registered with them (i.e., paying the fee), you can’t freely distribute Windows or macOS software. Apps from unidentified developers will display a message saying the app is damaged or blocked and can’t be opened.

To bypass this, open the Terminal and run one of the commands below (depending on your operating system), replacing <path-to-app> with the correct path to where you’ve installed the app:

• Windows: Unblock-File -Path <path-to-app>
• macOS: xattr -d com.apple.quarantine <path-to-app>

To build this project, you will need the following dependencies installed in your computer:

• Golang
• Task

With all the dependencies installed, in the project's root folder run the command:

$ task cli os=<operating-system> arch=<architecture>

Where:

• <operating-system>: can be windows, darwin (macOS), or linux.
• <architecture>: can be amd64 or arm64.

For example, if I wanted to build the CLI for Windows, on architecture AMD64, I would run the command:

$ task cli os=windows arch=amd64

mediaorient is released under the Apache 2.0 License. See LICENSE for details.

Vinicius Egidio (vinicius.io)