Digital photography leads to a veritable flood of images – and over time, unfortunately, to an uncontrolled chaos on the PC. Thanks to Artificial Intelligence, even tens of thousands of recordings can be automatically analyzed, tagged and sorted in a short time. The result is amazing!
How to manage photos perfectly
© © Carmen Steiner – Adobe Stock
Each individual shot does not cost an extra cent and is created with your smartphone anytime, anywhere: a huge plus point of digital photography compared to analogue films and paper prints of earlier days. At the same time, however, this leads to a flood of images and, based on experience, a veritable chaos of images.
There the files are spread over some directories, data carriers and devices – preferably double and triple. After years, you no longer know exactly what is secured where. And because the photos are neither automatically named nor given keywords, the only thing left to do when searching is the visual view. Please imagine that you would save your text files according to the pattern “Word12345.doc” and would not have a keyword search! The situation with photos is exactly the same.
There are also ways for images to resolve this chaos and bring order to the collection – thanks to artificial intelligence, it even works automatically. It suddenly becomes fun to look at your own recordings again, for example because you can use the keyword search for a specific person at a specific location in a specific year to immediately see the appropriate recordings.
Fake photos, videos and texts
Merge all photos in one place
External USB hard drives in 2.5-inch format are handy and available with a capacity of one TB from 40 euros: enough space for almost every private photo collection.
Of course, we do not know on which USB sticks, memory cards, other data carriers, possibly worn-out smartphones, computers and so on. In any case, it makes sense to bring them together in a central location. This can be an internal, external or network hard drive.
The following model calculation gives an order of magnitude for the space required: If you take 1000 pictures with a file size of five MB each year, the space adds up to 5000 MB or five GB per year. When processing the photos from ten years, it is 50 GB. The amount of data will of course differ from you, but even with numerous duplicates, everything fits on an external 1 TB USB disk for the price of a good 40 euros. You will not be spared the manual copying via Windows Explorer, but you do not have to make any particular effort with regard to the organization and naming of the directories.
Delete duplicates to free up space and improve order
Both QNAP (back) and Synology (front) provide software for your network hard drives that automatically sorts and tags images.
© Synology & Qnap
In the next step you will feel
duplicate photos and delete unnecessary duplicates. The program is not limited to comparing the file name, size and date, but also analyzes the image content and assigns a value to each picture. If the values of two photos differ below a threshold, the tool classifies them as duplicates.
That’s how it’s done:
After installation, start Similar Images via the context menu (right mouse button) with “Run as administrator” and change the language to German using “Languages -> German”. In the “Search” tab, use the preset “Normal search”, in the second tab “Semi-automatic deletion” change the deletion criteria in the following order (from above) “Smaller file – Lower resolution – Older file” and activate the option “0 -File files automatically “. With this, Similar Images automatically deletes duplicates according to the criteria listed. Click on the folder with the plus sign in the center right under the “Comparison directories” window, select the top level of your central photo directory and start the image analysis using the magnifying glass icon.
With small picture collections you can still delete the duplicate photos manually, with thousands of pictures you can let Similar Images do it better for you automatically.
With our test sample of a good 13,000 photos (approx. 60 GB) on a medium-speed PC, this took around 15 minutes. Depending on the similarity and number of recordings as well as the set comparison threshold, you must then manually decide for some duplicates whether and which of the duplicates should be deleted. At the end the protocol shows the result: With our photo set there are a good 1000 “similar image pairs”, which frees up more than four GB of hard disk space.
Automatic motif recognition versus manual keywording
If the corresponding details of the compass and focal length are saved in the Exif data of a picture, Geosetter shows the direction and angle of the view in addition to the location.
With the central storage location and the deletion of duplicates, there is already a bit more order in the image collection, but nothing more. Anyone who summarizes their files in folders such as “Christmas 2017 at…” or “Pentecost 2012 London” already has a slight idea of the content. However, you still don’t know which motifs are in the individual shots. To do this, the photos have to be tagged first – and that would take quite a long time with many thousands of files.
But there is a remedy, the IT corporations from the USA analyze the photos with artificial intelligence, recognize motifs, faces and locations and independently assign the appropriate search terms to them. Before we do this at Amazon and Google with our image sample, it should be noted that commercial photo agencies like
perform similar tasks and automatically assign objects and faces. There are only a few freeware with this function, an example is
. Later more.
But first to Amazon and Google, both of which offer cloud storage for storing photos. Amazon generally provides five GB, Prime members even unlimited space for private images in full resolution. Google also does not limit the storage for private photos, but automatically reduces their resolution to 16 megapixels when uploaded. There are exceptions only for owners of pixel smartphones 1, 2 and 3 – limited to January 2022 for the 3-series model and until the end of 2022 for the 2-series model.
In the following we show how you can upload your photos to the Amazon and Google cloud on a Windows PC and then analyze and automatically tag them.
Amazon Prime: Upload images to the cloud for analysis and tagging
Uploading photos to the cloud using Amazon’s Photos software is easy, but takes time. Then the photos can be automatically tagged.
Install Amazon’s Photos software and log in to your account. To upload your photos, click on the blue “Browse” button in the central area “Move photos by drag & drop”. Here you select the main directory of your central collection and confirm this with “Select folder”. In the next window, change the default setting for “What should be saved” to “Photos” (or “Photos and Videos”), activate “Avoid duplicates” below and start the upload with “Save”. This process can take time, after all uploading most Internet connections is much slower than downloading. Ten GB or around 2000 photos take more than two hours even with a fast VDSL-50 connection.
Unlike cloud storage, automatic motif recognition on Amazon only works with Prime membership for € 69 a year. Log in to the browser
on the photos page
, click on your account in the top right corner, select “Properties” and activate the “Find people, places and things” function. The image analysis starts in the background.
The result is really impressive: in the image overview you will not only find your pictures, but also the “People” and “Places” functions on the left. A mouse click on “People” suggests – of course only for pictures with people – face sections to which you can assign names via “Who is that?”. If the same person is summarized separately again, confirm the assignment again with the same name. With smartphone photos, Amazon automatically assigns each location to its location: not “cryptically” with degrees of longitude and latitude, but with the location name. The same applies to the recognition of motifs, naturally dependent on the image content.
This keyword search can now be combined almost arbitrarily. Here is an example: If you type in the Photos search bar “Person A”, “Person B”, “Location XY”, “Subject XY”, you will really only see the pictures that match these search terms.
summarizes the most important things.
In addition to the location, Geosetter also shows the direction and zoom section when taking pictures of smartphones and cameras equipped with a compass. This makes motivation identification much easier in some cases.
The most popular free image viewer for download
Google enables automatic image analysis for everyone
After years, you often no longer know exactly what is shown in the photos. The automatic assignment of the recording locations also helps here – shown here using Google as an example.
At Google, uploading the photos to the cloud and image analysis work in a similar way, but unlike Amazon, automatic tagging is open to everyone here.
That’s how it’s done:
NewsABC.net, install and start the Google software
Backup & sync
on your PC, log in to your account. Accept the presetting “Back up photos and videos” and in the next step define your photo directory via “Select folder”. Depending on the amount of images, you decide for the unlimited storage space (“high quality”) or the “original quality”. The upload starts with “Start”.
To be able to use the automatic image analysis, switch on the “Face grouping” in the Photos app on the smartphone via “Settings -> Group similar faces” – this setting does not exist in the browser on the PC. Since the assignment of people may not be active immediately, please be patient until content can be seen in the “Albums” view in the browser or in the smartphone app in the “People” folder. By clicking on a recognized person, you assign them a name similar to Amazon. Like Google, Amazon makes it possible to combine several search terms for people, locations and automatically recognized image content. Motif analysis works well even with videos.
A typical case that illustrates the difference in image recognition. While the search engine Yandex correctly locates this image in Israel, Google shows everything.
Another tip for determining the location, which in addition to the GPS information and – if activated – the location history also recognizes attractions: Google does not list all of the recognized locations in the “locations” by far. To search specifically for photos of a location not listed here, go back to the “Photos” overall view and type in the location in the search bar at the top.
Two aspects apply equally to Amazon and Google for image analysis in the cloud: Both services currently do not offer a correction option for changing incorrectly assigned tags – apparently they think their AI is infallible. In addition, both providers are granted a number of usage rights by uploading the images to the cloud. It is difficult to estimate how far these actually go. While Amazon does this specifically
documented for his photos service
, Google leaves it
Automatically save and analyze all new photos
The digiKam PC software does not have automatic tag allocation for image motifs, but nevertheless offers many useful filter functions, including “Search by area”.
Now that your existing images have been sorted and tagged, you can extend this to all new ones. In addition, many cloud storage services offer the option of uploading photos taken with the smartphone to the Internet and saving them there – either immediately or only in the WLAN.
At Amazon and Google, image analysis starts at the same time, pure cloud storage like Dropbox only synchronize new photos with your PC in this way. A synchronization tool such as takes over from the storage folder on the hard disk
the recordings then in your central photo directory, which digiKam or another management program can access. Pure Sync is also suitable for storing photos from digital cameras or memory cards centrally on the computer or network hard drive. From there, in turn, Amazon Photos and Google Backup & Sync upload new content for analysis into the cloud and thus into your photo collection.
If you save your photos on a network hard drive (NAS), you can also use the PC and smartphone apps from NAS manufacturers. QNAP and Synology put in your application
or moments now also intelligent image recognition, which automatically indexes and sorts photos by people, places and topics. If you have a drive from one of these manufacturers, you should definitely try out the apps on the NAS and the versions for mobile devices (Android and iOS).
Motif recognition over the Internet
A holiday picture taken a long time ago, an unidentifiable motorcycle model on the side of the road, or the defective washing machine where the nameplate is so faded that you cannot order a spare part: three examples that underline the meaning and benefits of automatic image and motif analysis.
and various other search engines can upload and scan photos. The best results in our tests were achieved by the Russian-Dutch search engine
on their picture search.
Google’s Photos app offers with the function
the ability to analyze objects in images. In this way, many objects, buildings, animals, plants and much more can be determined.
digiKam: Sort and tag recordings offline
DigiKam’s “Advanced Search” offers a wide range of filter options, including narrowing down to photos taken at a certain height above the sea.
Free software to automatically sort and tag images is rare,
is an exception. However, digiKam does not work as easily, comfortably and exactly as the online services. This is not surprising, since the software runs offline on the PC and does not use any artificial intelligence in the cloud – with the advantage that no one else scans, sees and evaluates your images. The facial analysis of our 13,000 test images on a PC with a seventh generation Core i5 CPU alone took almost two hours. In addition, a considerable amount of correction was required afterwards because digiKam did not always correctly assign pictures to the same person.
That’s how it’s done:
After installation, use the setup wizard to define the directory with your photos, define a folder of your choice for the tag database and answer the question about saving the photo information in the metadata with “Add information to files”. This stores important data directly in the photo files themselves and not just in the database. Finally, it takes a moment for the program to display the pictures in the album view. If you ever “lose” the albums, make them visible again using the small, vertical “Albums” tab at the top of the toolbar on the left side of the screen.
First carry out the face recognition by clicking on “People -> Search collection for faces” at the bottom left. When the analysis is complete, assign the individual faces their names. The geographical limitation of recordings is also practical: To do this, click on “Map” on the left, followed by the first small symbol behind “Search in area”. You define the search area with the mouse using drag & drop on the map on the left. It is also useful to search by date and timeline; you will also find both in the function bar on the left.
In contrast, digiKam lacks intelligent motif recognition; So you cannot avoid manually assigning keywords here. To do this, select one or more images in the album view, right-click on the selection and select “Assign keyword -> Add new keyword”. Enter one or more terms in the text field, each separated by a space. You can also do the tagging in several rounds, for example by first adding the place name to all the pictures taken in one place and then adding the motifs of the individual pictures.
In the end, the note on how to filter and find photos using the given keywords should not be missing. To do this, type in the search term (s) using “Search” on the left of the program, and even more search options are available using the “Advanced Search” below.
All of this is only a small part of the many functions of digiKam; therefore you need some time to familiarize yourself with the operation.