Created an online service to remove the background image faster Photoshop
Neural networks can already perform with graphic images previously unimaginable things. For example, in July 2018 NVIDIA demonstrated the algorithm that eliminates noise from photos and extra labels. Unfortunately, ordinary users can not take advantage of his skills, but there are developers who generously share their projects with ordinary people. One of them was the service, allowing you to quickly remove the background from any picture with a man.
Service called Remove.bg - he was an excellent example of how machine learning-based projects become easy consumer tools. Algorithms that quickly remove background from photos already exist and are available in the public domain. However, unlike them, this service provides the algorithm in the most convenient form for use.
To remove a background image and leave only the main object, you simply need to upload a file to the site, or to insert a direct link to the image - the algorithm itself will do the work and issue a PNG-file with a transparent background. The tool works only with photos of people, so to remove the background from pictures of animals and objects will not work. At some point in the service copes with the removal of even the best Adobe Photoshop and similar programs. The difference is especially noticeable in photos of people with long hair - ANN accurately distinguishes them and adds transparency.
Of course, the meet and inaccuracies. The algorithm does not always accurately determine the boundaries and may remove part of the human face. For example, the image below visible remnants of the dark background. However, the service is able to help out in the moments when the hand is not Photoshop or other editor. The remaining defects can be corrected manually, and if you insert a PNG-file in a different background, ugliness at all can become invisible. .
Retrieved from the algorithm can be anyone on site Remove.bg. Service is completely free.
Do you like the service? My opinion of him, you can write in the comments or in our Telegram-chat.