Artificial intelligence is increasingly used in digital photography. One of the most used types of AI in photography is deep learning: it is a type of artificial intelligence derived from machine learning where the machine is able to learn by itself (thanks to data models), unlike programming where it simply executes predetermined rules.
This AI is already available in the photo mode of many high-end smartphones to allow them to automatically detect scenes and even extrapolate images.
AI applications in photography and video
Among the different applications of AI in photo and timelapse, we could mention:
- interpretation of images: analysis of their content (objects and their location) more or less thorough (different levels of analysis), up to the interaction of these objects.
- image generation with style transfer: www.deep-style.io
- photo creation from drawings: https://techcrunch.com/2019/03/18/nvidia-ai-turns-sketches-into-photorealistic-landscapes-in-seconds/
- generation of non existent faces (https://www.thispersondoesnotexist.com/)
- detection of “deep fakes” which are fake photos or videos and become viral on the internet. For example, the start-up Truepic wants to offer counterfeit detection services based on AI
- optimization of the shots in particular the “selfie” mode improved thanks to a bokeh effect which consists in the addition of depth of field during the detection of one or several faces
- automatic shooting when smiles are detected
- intelligent improvement of many images in a few seconds automatically, such as the application for professional photographers developed by Meero:: https://www.meero.com/technology
- intelligent setting for SLR cameras: https://www.kickstarter.com/projects/2092430307/arsenal-the-intelligent-camera-assistant-0?lang=fr
- manufacture of works imitating great artists paintings: https://www.futura-sciences.com/tech/actualites/informatique-ordinateur-fabrique-faux-rembrandt-y-meprendre-62366/
Artificial Intelligence & timelapse
Because the timelapse reveals the invisible, this technique is particularly subject to the use of artificial intelligence. Many features and objectives result from this and among them:
- detection of objects, people or animals through image recognition training and for counting, securing or blurring purposes.
- automatic and intelligent improvement of images: some tools such as Let’s Enhance allow to improve the resolution of images thanks to the SISR (single image super resolution) technology with Artificial Intelligence. You want to try ? It’s this way : https://letsenhance.io/
- smart prediction calculations for object/person tracking (source Tali Dekel):
- content creation : it is for example the transfer of style of an image: from photo it becomes drawing (https://deepart.io/) or the construction of a realistic photo from a very simple drawing and little detailed (https://techcrunch.com/2019/03/18/nvidia-ai-turns-sketches-into-photorealistic-landscapes-in-seconds/)
- automatic shooting according to certain scenarios: the Photobooth mode of the last smartphone of Google offers for example to automatically realize the best selfie
- selection of the best photos for the timelapse video generation thanks to intelligence algorithms specific to each application / tool.
- transformation of a standard video into slowmotion video thanks to the AI by interpolation of photos (source NVIDIA):
Some timelapse solutions already use artificial intelligence algorithms (https://enlaps.io/tikee-artificial-intelligence/?lang=en) for the automatic calculation of photo parameters according to the scene, object / vehicle / person recognition, compliance with the GDPR through the anonymization of snapshots and blurring people, improving videos, selecting the best shots for videos to perfect rendition, etc.
Did you like the article? Feel free to share it!