![]() The pre-trained recognition model in Digikam 7.0 also recognizes the portraits of pets, for example dogs. This method is now included in Digikam 7.0 and, according to the release notes, delivers a recognition rate of 97 percent at an acceptable speed thanks to multithreading.Īnother advantage is that the new detection is not limited to human faces. In a new project for the Google Summer of Code 2019, the breakthrough came with the deep neural network module of the newer OpenCV library. This method brought a higher recognition rate, but it was too slow: the recognition of faces in images with 112 by 92 pixels took around 8 seconds, the developers wrote. Further runs then include these patterns.Īs part of the “Google Summer of Code”, Digikam began to use facial recognition via neural networks and with students in 2017 the C ++ library Dlib to improve. ![]() The new face recognition from DNN OpenCV in action: Digikam 7.0.0 recognizes portraits of people as well as animals independently and then suggests the manual categorization of this image material by name. What sounds like an impressively high rate is too little for everyday use, as the user feedback in the bug tracker showed. The algorithms were based on “Cascade Classifiers” from Intel’s OpenCV library, remained largely unchanged up to the current version 7.0 and left something to be desired in their precision: The Digikam developers got around 80 percent recognition rate out of OpenCV. Since Digikam 2.0, photo management has included algorithms for recognizing and automatically tagging people in pictures. The most important innovation that emerged as the highlight of this version at the first beta last December is a new face recognition. Face detection: more reliable with neural networks Once the program has been resolved, the new development pipeline should also be expanded to CD and deliver finished Digikam builds faster. According to Gilles Caulier, one of the main developers of the project, the GitLab instance of the KDE team at the CI can currently cause difficulties in compiling. The move of KDEs and its applications was in preparation for a few months and was completed by Gitlab with a Activity summary celebrated.įor Digikam developers, the scope of Digikam source code on Gitlab poses a number of hurdles. For KDE’s close sub-projects, the various individual tools previously used became a lot of work and Gitlab now summarizes a lot. Digikam is the first version of photo management to be found on the KDE project website at Gitlab. The developers have completely moved the source code management like the other central KDE projects from various individual tools such as Phabricator and Bugzilla to Gitlab. Digikam 7.0 is now also available as a flat package for Linux systems. After almost 20 years of development, Digikam is no longer just a figurehead of the KDE working environment under Linux, but is also available as a port for Windows and Mac OS X. While in beta stage, Tonfotos is available for free to the beta program participants.During this delay it was a matter of fixing problems that occurred afterwards when building the various binary packages. Quickly find what you need regardless where files are stored - on your computer, external drives or NAS. Tonfotos simplifies browsing of large photo collection by grouping shots by events, dates, people, locations and so on. Tonfotos (Freeware, still in Beta phase) It is able to load images on the internet from a page or image URL. Sequential can display folders and archives (ZIP, RAR, CBZ and CBR) of images (including JPEG, PNG, and GIF) and PDF files. Sequential (BSD licensed), see: GitHubĪn image viewer for Mac OS X originally designed for opening a folder of images and displaying them in order. The images original size fluently with very short load times and no Galapix, C++ implementation, see: GitHubĪn image viewer that allows you to directly zoom into largeĬollection of tens of thousand of images from tiny thumbnails down to Quickly sketch an image or click on an existing photo to find other photos containing similar images. ![]() ImgSeek - Intelligent image database, see: sourceforgeĪlthough it's a full-featured image viewer and manager, this app focuses on enabling content-based search. (Can you say “new baby?”) Fotobounce detects the faces, suggests names which you approve and voila! The photos are tagged and easily searched – especially helpful for putting together family photo montages for anniversaries, birthday parties and more. This is a great tool when you have hundreds, if not thousands of photos.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |