My Workflow for Efficient Keywording of a Stock Photo

Keywording determines the fate of the stock photos. I want to explain this important process with a sample photo of a cat.

First of all, there is no point in tagging a photo of cat with the keyword “cat” only, because when you search for “cat” on iStock, you are seeing 457000 results today. So if you cannot find any other interesting keywords to tag a cat’s photo, do not waste your time by uploading the image.

Supposing that you want to upload the image in any way. In that case, you can make some adjustments on the image. For example: changing colour of only one eye of the cat may be an option to add new keywords and separate it from other cats.

But this small change does not remove the necessity from adding keywords “cat”, “animal”, “outdoor” etc. for the image. I hate dealing with things that are far below my abilities. Isn’t it a boring routine to look a photo, see the cat and write the keyword “cat”? Do you know that there are applications that do this routine for you and generate keywords automatically? My motto as an engineer is “if something can be done by machines, it’s inefficient to do it manually.”

Here is my workflow of keywording a photo:

  1. I drag and drop the image on Keywordsready and let the tool generates keywords for me. I copy and paste the keywords to where I want to upload the image. (Keywordsready.com is a completely free keywording tool that uses artificial intelligence technology.)
  2. I ask myself the following question “What is in this image different from others?” and add a few more keywords that will help my image to separate from others. For this example, after editing eye of the cat: “heterochromia”, “variety”, “colourful”, “anomaly” etc.
  3. I ask myself the following question “if I were a buyer, which keywords would I use to find my image?” and add a few strong keywords which are conceptual generally.

Sometimes I skip the steps 2 and 3 because Keywordsready might have generated all necessary keywords for me.

An example how Keywordsready can increase your productivity

Assuming that you are a travel photographer and you have a Canon 5D Mark IV. Last month you traveled a few countries and shoot hundreds of images of streets, natural landmarks, buildings, people etc. Now you are at home and want to upload your images but you think that keywording is an annoying work because you have to remember where the images have been taken and name of the landmarks. Don’t worry! If GPS of your camera is turned on during your travel, leave the rest to Keywordsready that can find not only country, city, district of the image but also name of the landmarks. Although only this feature is enough to increase your efficiency, there are more features like finding number and ages of people visible in the image, technical properties of the image and many more…

Can Artificial Intelligence Create Fashion Designs?

Google and online store Zalando’s joint project named Project Muze is the first attempt in starting a period of digital fashion design. This project uses visual AI of Google called DeepDream which was used to create psychedelic works of art before. Project Muze’s claim is to make users their own muses and create designs according to their likings and interests. The neural network used to analyze the design options is trained by the style choices of 600 trendsetters.

We need to be aware of one detail here: The claim of this project is to take ‘inspiration’ from your choices and make designs by using you as a muse. However, it does not claim to create products that you will use in daily life. They also accept that Project Muze is still in early stages of trial period.

Black-and-White Photos Are Colorized By Artificial Intelligence

In a nutshell, Deep Learning means the AI’s ability to learn analyzing and identifying data. This turned into an algorithm that colorizes black-and-white photographs in the hands of a startup called Algorithmia.

Deep learning represents computer software’s gaining the ability to learn by imitating the human brain. Seattle based startup Algorithmia currently have 15 different deep learning algorithms put into the service of developers. Among these algorithms are the ones that can recognize faces in a crowd, detect nudity, colorize photographs and sense the stress level of a person by the tone of voice. Algorithmia team plans to offer their products to application development companies that work on this sector and need artificial intelligence software. The company’s web based photo colorization tool is quite interesting.

You can get to their website through this link and copy/paste the URL of the photo you want to colorize. The artificial intelligence then will load the result and show you both versions of the photo together.

After colorizing the black-and-white photos you chose, it is possible that you may not like the results very much. However, the impressive thing here is that the AI learned how to colorize objects, sceneries, people or animals by taking one million different photos as examples. It is impossible to deny the fact that artificial intelligence and deep learning will affect our lives even in the short term.

Watch The First Movie Written Completely By Artificial Intelligence

Creativity is a quite critical threshold for artificial intelligence. Machines make constant progress each day by learning how to paint or compose music and now they turned their eyes to movies. First attempt was done with a short movie called Sunspring and it is known to be the first movie written by AI.

The movie was released through Ars Technica and has Thomas Middleditch of TV series Silicon Valley starring. If you have never heard the sayings of an AI bot before, it is better if you get used to them before watching the movie. Because the dialogues in the movie may be hard to follow for beginners.

Sunspring is categorized as a science fiction movie and this is befitting when you consider the dialogues, costumes and scenery. However, the main reason is that the software which wrote the script is trained with the science fiction movies of 80s and 90s. Ross Godwin, who is a researcher on Artificial Intelligence in New York University, is the developer of this software called Benjamin. Together with the director of the movie Oscar Sharp, these two have been working on the idea of machines that can write screenplays. So, they have developed a neural network called Long Short-Term Memory (LSTM) as a solution to this problem.

However, it is noted that turning a screenplay into a movie is not an easy task. Sharp states that Benjamin’s script has odd and incomprehensible sentences like the one where it says the character is standing up on the stars and sitting on the floor at the same time. He also adds that the movie team first laughed at the screenplay and had to reinterpret it. The actors also have key roles in translating the script as they are the ones who will act it out to the audience. It is possible to think that years from now, the machines will be making movies among themselves for each other and these movies will only be understandable by these machines.

AI Technology Can Create Images That Look Real To Humans

Artificial intelligence is the most innovative and different area of the technology world so it became the leading sector tech giants invest in.

Innovations like Microsoft’s TwinsOrNot website, Facebook’s Moments application and Google’s system that automatically adds descriptions to photographs, show us the direction of AI evolution. Facebook’s latest study is one of these.

Latest study Facebook research team announced was on image recognition and generation. The AI robot that was developed is capable of analyzing various different objects (anything), learning what they look like and creating new photographs accordingly. This is actually based on the same logic human brain perceives objects. Meaning that our brains first analyzes an object to learn what it looks like and then when it sees a similar object, it recognizes that this new object looks like the previous one based on the information gathered. This system creates photographs based on the same method.

In this experiment, Facebook’s AI robot generates 64×64 pixel life-like images. 40% of the test participants believe that these images are real and not created by an artificial intelligence robot.

The AI robot works in two different steps: First, the artificial neural network generates a photograph based on a random vector. Then, the realism in this generated photo is analyzed. In time, it is planned to make this robot capable of creating photographs in larger scales.

It is currently unknown what Facebook will do with this new technology or which of their services will benefit from it. However, the company showed its obvious intentions on implementing AI technology into Facebook services by recently announcing the app named Moments.

Face Recognition System that Picks Out One From a Billion

“Picking out one out of a billion” is not just an expression for the NTechlab team. It takes less than one second for the system to find a person in a database of one billion photographs. Artem Kukharenko says: “We are the first ones to learn how to use large image databases efficiently.” He adds: “This plays a key role in solving real-life problems like finding a criminal in real-time or identifying a regular customer from store security cameras.”

Asides from the perfect database search function, the algorithm’s recognition accuracy is also quite high. The secret of these abilities are deep learning and neural network architecture. The laboratory states that it is crucial to understand how the accuracy rate is measured. Usually the accuracy rates are lower for larger image databases. Finding a person in a million is much harder than finding him in a hundred. In MegaFace Benchmark Championship, NTechlab reached 73% accuracy rate in a database of one million and 95% accuracy rate in a database of 10.000 images. For the comparison of two images with the intent of verification, the system works perfectly with 99% rate.

The heart and the brain of the system is neural network. The most difficult tasks AI systems encounter are the ones human brain can do with ease. Recognizing a human face in a crowd or identifying the gender of a dog are easy for us. Our decisions are determined by multiple factors and previous experiences. The neural networks used in the system are based on the same model. Various mixed signals are sent to the neuron. According to these signals, an output signal is formed. If the system makes an error, the formula tasked with weighting the input signals is corrected. Learning from the mistakes, raises the accuracy of the system. The interesting thing is that our brain makes the same calculations with a machine when we see a familiar face. The system uploads a unique image like the photo of a random passer-by to FindFace. First of all, the face in the photograph is identified instantly. Artem Kukharenko states that even if it seems weird, the process that is the most resource intensive is the face-recognition step. The team is now working on speeding up the algorithm and making it less resource intensive.

In the secong stage of recognition, it is necessary to build a feature vector through using a trained neural network. The feature vector consists of 80 numbers that contain information about the face. The numbers may be similar for the same person but can vary greatly for two different people. The search system is based on these differences. At this step of the recognition, the system must identify the constant information even under conditions like the person wearing glasses, having a beard or if there is a time gap of long years between two images.

The last stage involves searching the image database for this “face.” NTechlab development team needed efficient workstations for carrying out many studies and making calculations in order to build a search algorithm. Intel® Core™ i7 processors were chosen to help HTechlab design a truly efficient tool to make efficient searches of a large number of images. This way, if the database size is increased 10 times, the search time will only increase by 1,5x. The system offers a wide range of application choices from dating apps to government security systems. Products developed by NTechlab stand out in the security area, which is considered to be crucial. Existing systems offer solutions for photo comparison problems such as comparing a passport photograph to the images in an airport database. However, these systems are not sufficient enough to establish safety in large scales like for a whole town. NTechlab’s solution can process feed from thousands of surveillance cameras in real-time and detect criminals in a big city.

In retail sector, it is possible to use this system as a replacement to discount cards. It can easily record the feature vector of the customer by using his photograph. Later when the same customer enters the store, the security camera can identify the customer’s face. This method conveniently solves the problem of storing personal data and makes it unnecessary for the stores to keep information like name/last name and phone numbers. Entertainment sector has already implemented systems that can make visitor searches against photograph databases. The visitors send their selfies to the robot and the robot brings up all the photos these visitors are in. This system is quite useful in festivals, weddings and other similar activities. According to Artem, this system is currently being used with success in Alfa Future People Festival and a theme park in Australia.

NTechlab does not plan on settling with this success and they have already put a cloud based facial recognition system into service. Companies can upload a photograph database to the system and run searches in it. They can also implement this into their services. Soon, an SDK (software development kit) will be made available for third party developers and in a few months after that, a facial recognition system for factory security purposes will be online.

Currently, the lab is working on optimizing the algorithm and improving its accuracy rate. They are aiming to develop a module that will recognize the emotions in a photograph and distinguish between a real person and a photograph. In order for artificial intelligence to succeed, there needs to be continuous development.

Artificial Intelligence Will Recognize Fonts

Adobe, creator of Photoshop, is preparing to release an add-on that will help the users

Adobe officials said that their AI called DeepFont will work like the song identification app Shazam.

The main function of DeepFont AI will be to detect and identify the fonts in the designs users come across in daily life.

Adobe announced that unlike any other font recognition systems currently put into the service of the users, DeepFont will have the ability to learn.

Users curious about the fonts they see, will upload the photos they take to DeepFont and the AI will check the style in the photo againts the 20.000 fonts in the database to come up with 5 results.

Chief of Data Processing in Adobe Anil Kamath says: “To use this application, all you have to do is to take a photo with your phone.”

Facebook Image Search Recognizes Objects In Photos

With Facebook’s new image search engine, users are able make searches according to the objects in photos. For example, a search for “photos of me in a blue T-shirt” will bring all the photos of the user in which he wears a blue T-shirt.

This new function seems to be revolutionary for the internet world. Searching photos over the internet according to “content” will become widespread. It will be possible to make Google image searches like “green car with a female driver.”

This feature can also be used for detecting and removing photos containing violence and pornography.

Artificial Intelligence that Identifies Criminals Based on Facial Recognition

Scientists Xiaolin Wu and Xi Zhang of the Shanghai Jiao Tong University in China, developed a system that can identify criminals through facial recognition. They did that by adding artificial intelligence and neural network technologies to the hypothesis which was put forward in 19th century.

The AI which is equipped with several artificial recognition algorithms, examined photographs of guilty and innocent people.

In order to test the neural network, scientists used 1856 ID photos of shaved men aged between 15 and 56. Half of these photos belonged to men with criminal records. 90% of these photos were used to train the AI and the remaining 10% was used to test it.

The results were simply amazing; it was seen that the neural network detected criminals with a 89,5% accuracy rate.

Facebook’s Photo Description Tool For The Visually Impaired

According to Facebook, there are nearly 300 million visually impaired people around the world. If these people use smart phones, they will be able to hear the explanation of the image on the screen. Before the update in the operating system, the smart phones only voiced the “photograph” word and who shared it.

In the first stage, the photo descriptions will be limited to 100 words so the users will not hear much detail. Visually impaired users will hear a sentence like “there are three people, smiling, outside” but the AI will not say that these people are also eating and drinking.

Facebook is taking things slow so as to avoid any possible embarrassment. Last year, Google’s search engine identified a black couple as gorillas and the company had to apologize for that.

People at Facebook say that they will improve their technology in time and it will be possible to make more precise descriptions. They are even aiming to answer the user questions about the images.

In the announcement, they say: “Each step taken in object recognition technology will increase accessibility so that we will reach more people. When people connect to each other, we will be able to achieve extraordinary things not only personally but also as a society.”

For now, the first version of the program is limited to phones with IOS. The next step will be bringing this technology to Android phones. The company is planning to implement this technology to internet browsers in near future. Not only on the website but also through different applications, the estimated number of photos shared daily is 2 billion.

A New AI algorithm Can Detect Eye Diseases By Examining a Clean Image Of The Eye

Google has been working on important AI projects lately and according to a recent announcement, they have implemented a new algorithm to detect the eye disease “diabetic retinopathy” which can even cause blindness.

Research has been started in the medical world to determine the success rate of this algorithm. However, the doctors do not rush to trust an AI like this yet. If the algorithm improves and its accuracy rate rises, it will be one of the most important tools that can be used in the medical world.