Picture Pile Platform

In recent years artificial intelligence (AI) has made tremendous progress in many different fields due to improved machine learning algorithms, advances in computing power and the availability of big data. With machine learning platforms like TensorFlow (Google), PyTorch (Facebook), Azure (Microsoft) or CoreML (Apple), it is now possible for both researchers and application developers to create and use AI, which is often able to solve problems faster, more reliably and more cost effectively than humans. While there are many image databases such as Imagenet, CIFAR10 or MNIST available for training machine learning models to do computer vision tasks on some main categories, a significant problem today is the lack of image databases for specific classes to solve concrete problems. 

In recent years artificial intelligence (AI) has made tremendous progress in many different fields due to improved machine learning algorithms, advances in computing power and the availability of big data. With machine learning platforms like TensorFlow (Google), PyTorch (Facebook), Azure (Microsoft) or CoreML (Apple), it is now possible for both researchers and application developers to create and use AI, which is often able to solve problems faster, more reliably and more cost effectively than humans. While there are many image databases such as Imagenet, CIFAR10 or MNIST available for training machine learning models to do computer vision tasks on some main categories, a significant problem today is the lack of image databases for specific classes to solve concrete problems. 

Obtaining these data is a cumbersome process, as there is currently no platform that offers an easy set-up and the running of image classification campaigns in a streamlined, standardized, efficient and engaging approach, where the crowdsourced data are then open and free to use by anyone. The value proposition of the ERC PoC project Picture Pile is an innovative, commercially self-sustaining platform that uses the crowdsourcing game Picture Pile for efficiently and intuitively classifying images for machine learning. It will offer a standardized, easy-to-use way for everyone to freely and quickly set-up their own Picture Pile campaigns for collecting image classifications by the crowd in a gamified environment, where participation is incentivized through extrinsic and intrinsic motivations. After a pile of images has been sorted, the image classifications will be made publicly available. 

In addition, Picture Pile will provide a number of mechanics for guaranteeing the quality of the data collected such as collecting multiple classifications per image by different users or by showing training images with instructions. A number of premium services will be added to the Picture Pile platform, which will make the platform self-sustaining. The campaign creators will, for example, be able to pay the crowd (with a small share paid to the Picture Pile platform) for sorting the images. There will also be the possibility to pay for advertisements on our social media channels and for a prominent place on the main Picture Pile page to attract more users to sort their piles. In addition, for a small fee, the users will be able to use the Picture Pile Cloud to do specific computer vision tasks using the machine learning models that have been automatically built and trained with the Picture Pile data sets. With these new features, the Picture Pile platform has the potential to become a self-sustaining platform and the hub for crowdsourcing image classifications for AI and machine learning.