R&D Projects.
We evolve with the advancement of technology

ESA Space App Camp is one of the most interesting recurring international hackathons, organized by the European Space Agency - Centre for Earth Observation, based in Frascati near Rome. Norbsoft could not miss it. One of our representatives participated in one of the editions of this event.

The main theme of the event is the use of satellite data in mobile applications, acquired under the Copernicus program.

As part of an international, multidisciplinary team, a prototype of a farmer's assistant application based on a chatbot interface was prepared. After marking the location of their crops on a map, users could ask questions, both textually and vocally, and obtain necessary information.

For the conversational interface, Google Dialogflow was used, providing tools and algorithms for Natural Language Understanding (NLU), as well as the built-in Speech Recognition and Text-to-speech mechanisms in the Android system.

One of the functions was also the recognition of potential plant diseases based on photos of leaves taken using the application. The appropriate AI model was prepared using Tensor Flow with the Keras framework and then run directly on an Android phone using the TensorFlow Lite library.

The model was based on the Plant Village dataset, consisting of over 50,000 images of healthy and diseased plant leaves.

Information derived from satellite data was obtained using REST API and mobile Software Development Kit (SDK) Ramani, prepared by the Dutch Ujuizi Labs.

The Canadian agency Mosaic approached Norbsoft with the need to create a promotional mobile application in a short time for a campaign associated with the launch of a new type of Samsung TV. The Frame TV is a designer-looking device that, in addition to typical smart TV functions, also allows for the display of artworks or personal photos when the device is turned off, and the user is nearby.

The idea of the promotional application was to enable users to recreate their own photos using the style of famous painters, inspired by the extremely popular application at that time, Prisma.

In less than two weeks, a solution was created that allows for the collection of user contact information, lets the user take a photo, and send it to the server, where, after being processed by specialized algorithms, it was sent back as an attachment to an email.

Using the PyTorch library, a neural network for image style transfer was created, which was then trained on Vincent van Gogh's The Starry Night using the MS COCO dataset (Microsoft Common Objects in Context).

The styling operation requires a tremendous amount of computational power, which is lacking in mobile devices and even standard servers. As part of the implementation, Norbsoft used Amazon's cloud with instances equipped with GPUs (Elastic Compute Cloud g2.2xlarge).

Apart from the development work planned and commissioned by our client, Empik Foto Sp. z o.o., the Norbsoft team involved in the project conducted its own independent experiments aimed at analyzing the technological possibilities of implementing new, unique features into the product.

The Empik Foto application allows ordering physical products, personalized using the user's photos. Among the other products that are offered, you can find mugs.

Using the capabilities of Apple ARKit, we prepared a POC (proof of concept) for visualizing a mug with the user's photo in augmented reality.

In 2012, representatives from Norbsoft participated in the Isobar Create London hackathon. During a weekend spent in the capital of Albion, a prototype of a football collectible card game utilizing NFC technology was prepared.

Physical cards representing players contained tags that were recognized by a dedicated mobile application. With its help, duels could be fought according to the game's rules or trades could be conducted.

The application also contained elements of Augmented Reality, displaying additional animations when the camera was directed at a particular card.

In 2011, at the invitation of the mobile network operator Plus (Polkomtel Sp. z o.o.), Norbsoft participated in the 15th Polish Radio Science Picnic. At the network's stand, we presented the capabilities of Augmented Reality technology.

Using the Qualcomm AR Software Development Kit, we created an experimental implementation using virtual buttons - the mobile phone camera detected the covering of selected parts of predefined graphics. In our case, the image was a printed photo of a piano keyboard, and the buttons were specific keys. Pressing a key was associated with the application in the phone playing the corresponding sound.

creativeBox was an energy-efficient and cost-effective Raspberry Pi® microcomputer, operating under a system prepared by Norbsoft, which we additionally equipped with Bluetooth Low Energy and Wi-Fi radio communication modules. It allowed for the creation of solutions based on interactions with the mobile application in the user's device, which occurred only in a selected, physically accessible location inside buildings.

After being connected to power (standard charger with micro-USB connector) and the internet (wired Ethernet connection), we obtained a device that could work as a transmitter compatible with the Apple iBeacon standard (for example, informing the application that the user is near a point of interest inside a building).

creativeBox could also analyze Wi-Fi communication occurring nearby to detect devices on which the application was installed, with zero impact of this application on battery life.

creativeBox also had an HDMI output, allowing for visual communication with users via a connected screen (TV, projector) – for example, greeting a user in a salon within the creativeBox zone using their name, or displaying additional content that complements the information visible in a dedicated section of the application.

creativeBox could also operate in a limited hotspot mode – providing users with network access only to the extent that allows the application to function.

The device configuration was done through a cloud-based service.