Image processing is an integral part of the value chain for today's leading high-tech industries. Ever-increasing demands in algorithm robustness and accuracy are its main challenges. For specific applications automation reaches fundamental limits. However, even laymen excel at these problems.
Pallas Ludens resource-friendly harnesses this potential by enhancing cutting-edge low-level vision with individually optimized crowd-interfaces. This enables currently infeasible endeavors including ground truth generation for machine learning and performance analysis.
Pallas Ludens is a high tech startup which emerged from the HCI at the Heidelberg University, founded by image processing experts with decades of experience and strong connections to the scientific community all around the world. We developed a technology that not only allows us to solve even the most complex problems in image processing with crowdsourcing but also minimizes the amount of manual labor significantly through our award-winning processing methods.
Now, we are successfuly gamifying annotation tasks to further improve the efficiency of our offering. Our approach enables us to reach millions of gamers in all developed countries who are eager to quickly accomplish vision tasks in a matter of minutes at any time of the day.
Our team of experts is able to support any vision problem end-to-end, offering a full service as well as specialized support for each important element in the processing chain.
Read more about our solutions for Automotive Vision.
Collecting large amounts of data is a critical issue. The right sampling can significantly reduce engineering efforts or dramatically increase quality. We offer crowdsourced data collections for publicly accessible data such as outdoor environments. Our new technology for gamification of microtasks enables worldwide data mining by crowds.
During preprocessing our experts support the data homogenization and result specification. We will then tailor our annotation tools to obtain your desired results with maximum quality.
Based on your specific needs such as speed, quality, culture or time, image annotation is performed by a carefully selected subset of our crowd. We select individual crowdworkers based on a well-designed reputation scheme to guarantee fast and clean results. Each task can be presented to many crowdworkers for statistical evaluations.
During Postprocessing we carefully examine the individual results to ensure data quality based on eyeballing, statistical analysis as well as machine learning. All results are merged to a data package in the format you specified.
Our experts will help you with visualizations, machine learning and algorithm design based on the crowdsourcing results. If needed, we also connect you with vision experts in academia around the world for any questions we cannot answer right away.
"[...] (T)he product of the company Pallas LudensGmbH, made possible by its innovative technology, is highly in demand in many industries."
"[...] (I)ch(würde) die Firma gerne beauftragen [...]. So können wir in unserem Institut neue Modelle lernen und evaluieren und darüber hinaus auch neue Verfahren entwickeln, die bisher aufgrund der schlechten Datenlage schlicht und ergreifend nicht möglich waren."
"Die Minitasks von Pallas Ludens werden von Tausenden unserer Spieler genutzt, da sie die großartige Möglichkeit bieten, Premiuminhalte kostenlos zu nutzen."
"The technology developed by Pallas Ludens Computer Vision is most attractive for generation of such data bases as it is cheap, fast and scalable. Since the importance of image understanding and classification will further increase, we expect a high demand in the future."
Pallas Ludens GmbH
Im Bosseldorn 29
Tel.: +49 (62 21) 64 79 30 8
Tel.: +49 163 81 84 265
Dr. Daniel Kondermann
Tel.: +49 151 52 422 896