An interactive map that uses machine learning algorithms to detect fields and crops

OneSoil Map allows to explore and compare fields and crops in Europe and the United States (44 countries in total). The overview map helps to understand patterns of fields sizes and crops in different regions. Zooming in enables to know a specific field in detail: the hectarage, the crop, and the field score. Besides, the key feature of the map is that it allows users to see how these fields have changed over the past three years (2016 – 2018). The map reveals insights about local and global trends in crop production for farmers, advisers, and dealers. It helps to predict market performance at all levels and fosters smart decision-making.

Data collection and technology

The map was created by the startup OneSoil and is a continuation of the OneSoil digital farming platform, which automatically detects fields, identifies crops through satellite imagery analysis. The core technologies are based on AI, deep learning models, computer vision, IoT and original machine learning algorithms, which enable the company to process data in real time:

“First, we learned how to clean the satellite photos from artifacts to ensure correct processing of information. Second, we trained an algorithm to allocate field boundaries automatically. For the map, we simplified the boundaries so that the visualization is really fast. The accuracy of crop classification, or F1 score, is 0.91. Third, we trained another algorithm to automatically determine a crop that grows on a field. Fourth, we created what you can now see: the map.

 

 

Crotos: a project on visual artworks powered by Wikidata and Wikimedia Commons

Crotos is a search and display engine for visual artworks based on Wikidata and using Wikimedia Commons files.

The Wikidata extraction contains more than 133 866 artworks (September 2018) including 66 271 with HD image. This extraction is regularly automatically updated from Wikidata on the basis of the nature of the items and corresponds to visual artworks such as paintings, photographs, prints, illuminated manuscripts and much more.

The interface

Searches can be made by free or indexed search through a user interface. Results are displayed by chronological order with thumbnails. Links on thumbnails open a viewer with the image hosted on Wikimedia Commons.

It is possible to filter the results by type (painting, sculpture, print…) or to specify a period as a criterion.

By default, without criteria, a random selection is displayed. Besides with the Cosmos interface, it is possible to discover the artworks by indexation (par type d’œuvre, creator, movement, genre, collection…).

Descriptors

For each resulting image, the interface displays the title, the creator(s)  and the collection or the location where the artwork is maintained. These information are on Wikidata, a free, collaborative, multilingual, secondary database, collecting structured data to provide support for Wikipedia, Wikimedia Commons, the other wikis of the Wikimedia movement, and to anyone in the world.

Additional descriptors are date or period, nature of work, material used, inventory number, movement, genre, depicts, main subject, and so on. A full list of descriptors is mentioned here.

Contribution mode

The project has a contribution mode, useful for identifying missing information with facets. Finally, source is on github and the database of Crotos can be downloaded. Both are under Free Licence.

Sound heritage

Conserve the sound is an online museum for vanishing and endangered sounds. The sound of a dial telephone, a walkman, an analog typewriter, a pay phone, a 56k modem, a nuclear power plant or even a cell phone keypad are partially already gone or are about to disappear from our daily life.

Conserve the sound is a project form CHUNDERKSEN and is funded
by the Film & Medienstiftung NRW, Germany.

A visual catalog of material culture of Amsterdam

“Urban histories can be told in a thousand ways. The archaeological research project of the North/South metro line lends the River Amstel a voice in the historical portrayal of Amsterdam. The Amstel was once the vital artery, the central axis, of the city. Along the banks of the Amstel, at its mouth in the IJ, a small trading port originated about 800 years ago. At Damrak and Rokin in the city center, archaeologists had a chance to physically access the riverbed, thanks to the excavations for the massive infrastructure project of the North/South metro line between 2003 and 2012”.

Bellow the surface website presents the scope and methods of this research project, detailing the data processing and of approximately 700,000 finds. The website provides access to the photographs of more than 19,000 finds. The access to the complete dataset hasn’t been released yet, but a disclaimer note informs it will be shortly available.

In the main user interface (the object overview page), thumbnails of all 18.978 objects are arranged by estimated time of creation (period varies from 2005 AC to 119.000 BC). Users can scroll vertically along time with a mouse. A panel of facets filters the objects according to time range, object function (communication & exchange, personal artifacts & clothing, etc.), material (metal, synthetics, etc.) and location (Damrak or Rokin metro stations). The time range facet has an interesting feature: it is also a visual variable that shows patterns of distribution at a glance. The other facets indicate the objects occurrence through absolute numbers. Facets don’t require a preceding search and enable refinement (selecting an option of a facet changes the options occurrence in the other facets.

The main user interface

Selecting a thumbnail of an object reveals detailed information about it (close viewing). A bigger photography t is shown followed by detailed information about the object properties.

Close view of a selected object
Scrolling the page down reveals detailed information about the object

This research was conducted by the Department of Archaeology, Monuments and Archaeology (MenA), City of Amsterdam, in cooperation with the Chief Technology Office (CTO), City of Amsterdam.

Experiments by Google explore how machines understand artworks

Google Arts & Culture initiative promotes experiments at the crossroads of art and technology created by artists and creative coders. I selected two experiments that apply Machine Learning methods to detect objects in photographs and artworks and generate machine-based tags. These tags are then used to enhance accessibility and exploration of cultural collections.

Tags and Life Tags

These two demo experiments explore how computers read and tag artworks through a Machine Learning approach.

Tags: without the intervention of humans, keywords were generated by an algorithm also used in Google Photos, which analyzed the artworks by looking at the images without any metadata.

The user interface shows a list of tags (keywords) followed by its number of occurrence in the artwork collection. Selecting the tag ‘man’ reveals artworks containing what an intelligent machine understands to be a man. Hovering an artwork reveals other tags detected on that specific representation.

The user interface shows a list of tags (keywords) followed by its number of occurrence in the artwork collection. Selecting the tag ‘man’ reveals artworks containing what an intelligent machine understands to be a man. Hovering an artwork reveals other tags detected on that specific representation.

Life Tags: organizes over 4 million images from the Life magazine archives into an interactive interface that looks like an encyclopedia. The terms of the “encyclopedia” were generated by an algorithm based on a deep neural network used in Google photo search that has been trained on millions of images and labels to recognize categories for labels and pictures.

Labels were clustered into categories using a nearest neighbor algorithm, which finds related labels based on image feature vectors. Each image has multiple labels linked to the elements that are recognized. The full-size image viewer shows dotted lines revealing the objects detected by the computer.

The overall interface of Life Tags looks like an encyclopedia
Kitchen is a categoria clustering labels using a nearest neighbor algorithm.
Selecting a specific photo expands it and reveals the labels recognized by the machine.

Digital Humanities 2018: a selection of sessions I would like to attend

As Digital Humanities 2018 is approaching, I took a time to look at its program. Unfortunately, I didn’t have contributions to submit this year so I won’t attend the Conference. But I had the pleasure to be a reviewer this edition and I’ll also stay tuned on Twitter during the Conference!

My main topic of interest in Digital Humanities bridges the analysis of large-scale visual archives and graphical user interface to browse and make sense of them. So I selected the following contributions I would like to attend if I were at DH2018.

Workshop

Distant Viewing with Deep Learning: An Introduction to Analyzing Large Corpora of Images

by Taylor Arnold, Lauren Tilton (University of Richmond)

Taylor and Lauren coordinate the Distant Viewing, a Laboratory which develops computational techniques to analyze moving image culture on a large scale. Previously, they contributed on Photogrammar, a web-based platform for organizing, searching, and visualizing the 170,000 photographs. This project was first presented ad Digital Humanities 2016. (abstract here) and I’ve mentioned this work in my presentation at the HDRIO2018 (slides here, Portuguese only).

Panels
  • Beyond Image Search: Computer Vision in Western Art History, with Miriam Posner, Leonardo Impett, Peter Bell, Benoit Seguin and Bjorn Ommer;
  • Computer Vision in DH, with Lauren Tilton, Taylor Arnold, Thomas Smits, Melvin Wevers, Mark Williams, Lorenzo Torresani, Maksim Bolonkin, John Bell, Dimitrios Latsis;
  • Building Bridges With Interactive Visual Technologies, with Adeline Joffres, Rocio Ruiz Rodarte, Roberto Scopigno, George Bruseker, Anaïs Guillem, Marie Puren, Charles Riondet, Pierre Alliez, Franco Niccolucci

Paper session: Art History, Archives, Media

  • The (Digital) Space Between: Notes on Art History and Machine Vision Learning, by Benjamin Zweig (from Center for Advanced Study in the Visual Arts, National Gallery of Art);
  • Modeling the Fragmented Archive: A Missing Data Case Study from Provenance Research, by Matthew Lincoln and Sandra van Ginhoven (from Getty Research Institute);
  • Urban Art in a Digital Context: A Computer-Based Evaluation of Street Art and Graffiti Writing, by Sabine Lang and Björn Ommer (from Heidelberg Collaboratory for Image Processing);
  • Extracting and Aligning Artist Names in Digitized Art Historical Archives by Benoit Seguin, Lia Costiner, Isabella di Lenardo, Frédéric Kaplan (from EPFL, Switzerland);
  • Métodos digitales para el estudio de la fotografía compartida. Una aproximación distante a tres ciudades iberoamericanas en Instagram (by Gabriela Elisa Sued)
Paper session: Visual Narratives
  • Computational Analysis and Visual Stylometry of Comics using Convolutional Neural Networks, by Jochen Laubrock and David Dubray (from University of Potsdam, Germany);
  • Automated Genre and Author Distinction in Comics: Towards a Stylometry for Visual Narrative, by Alexander Dunst and Rita Hartel (from University of Paderborn, Germany);
  • Metadata Challenges to Discoverability in Children’s Picture Book Publishing: The Diverse BookFinder Intervention, by Kathi Inman Berens, Christina Bell (from Portland State University and Bates College, United States of America)
Poster sessions:
  • Chromatic Structure and Family Resemblance in Large Art Collections — Exemplary Quantification and Visualizations (by Loan Tran, Poshen Lee, Jevin West and Maximilian Schich);
  • Modeling the Genealogy of Imagetexts: Studying Images and Texts in Conjunction using Computational Methods (by Melvin Wevers, Thomas Smits and Leonardo Impett);
  • A Graphical User Interface for LDA Topic Modeling (by Steffen Pielström, Severin Simmler, Thorsten Vitt and Fotis Jannidis)

Artscope: a grid-based visualization for SFMOMA cultural collection

Existing projects in visualization-based interfaces (interfaces which enables navigation through visualization) for cultural collections usually focusses on making their content more accessible to specialists and the public.

Possibly one of the first attempts to explore new forms of knowledge discovery in cultural collections was SFMOMA ArtScope, developed by Stamen Design in 2007 (now decommissioned). The interface allows users to explore more than 6,000 artworks in a grid-based and zoomable visualization. Navigating the collection follows a visualization-based first paradigm which is mainly exploratory (although the interface enables navigation through keyword search, the visualization canvas is clearly protagonist). The artworks’ thumbnails are visually organized by when they were purchased by the museum. The user is able to pan the canvas by dragging it and the lens serves as a selection tool, which magnifies the selected work and reveals detailed information about the selected piece.

ArtScope is an attractive interface which offers the user an overview of the size and content of SFMOMA’s collection. However, the artworks in the canvas are only organized by time of acquisition, a not very informative feature for users (maybe just for the staff museum). Other dimensions (authorship, creation date, technique, subject, etc.) can’t either be filtered and visually organized in the structure of the canvas.

The video bellow illustrates the interface navigation:

Gugelmann Galaxy

Gugelmann Galaxy is an interactive demo by Mathias Bernhard exploring itens from the Gugelmann Collection, a group of 2336 works by the Schweizer Kleinmeister – Swiss 18th century masters. Gugelmann Galaxy is built on Three.js, a lightweight javascript library, allowing to create animated 3D visualizations in the browser using WebGL.

The images are grouped according to specific parameters that are automatically calculated by image analysis and text analysis from metadata. A high-dimensional space is then projected onto a 3D space, while preserving topological neighborhoods between images in the original space. More explanation about the dimensionality reduction can be read here.

The user interface allows four types of image arrangement: by color distribution, by technique, by description and by composition.  As the mouse hovers over the items, an info box with some metadata is displayed on the left. The user can also perform rotation, zooming, and panning.

The author wrote on his site:

The project renounces to come up with a rigid ontology and forcing the items to fit in premade categories. It rather sees clusters emerge from attributes contained in the images and texts themselves. Groupings can be derived but are not dictated.

 

My presentation at HDRio2018

During the paper session “Social networks and visualizations”, held on April 11 at HDRio2018 Congress, I presented the work “Perspectivas para integração do Design nas Humanidades Digitais frente ao desafio da análise de artefatos visuais”  (“Perspectives for integrating Design in Digital Humanities in the face of the challenge of visual artifacts analysis”).

In this work, I outline initial considerations of a broader and ongoing research that seeks to reflect on the contributions offered by the field of Design in the conception of a graphical user interface that, along with computer vision and machine learning technologies, support browsing and exploration of large collections of images.

I believe my contribution raises three main discussions for the field of Digital Humanities:

  1. The investigation of large collections of images (photographs, paintings, illustrations, videos, GIFs, etc.) using image recognition techniques through a Machine Learning approach;
  2. The valorization of texts and media produced on social networks as a valid source of cultural heritage for Digital Humanities studies;
  3. Integration of Design principles and methodologies (HCI and visualization techniques) in the development of tools to retrieve, explore and visualize large image collections.

Slides from this presentation can be accessed here (Portuguese only).

About machine capabilities versus human sensitivities

For Recognition, an artificial intelligence program that associates Tate’s masterpieces and news photographs provided by Reuters, there are visual or thematic similarities between the photo of a woman with a phrase on her face that reads #foratemer (out Temer) during a protest against a constitutional amendment known as PEC 55 and the portrait of an aristocrat man of the seventeenth century in costumes that denote his sovereignty and authority. In times when intelligent and thinking machines, like chatbots, are a topic widely discussed I wonder if the algorithms that created the dialogue between these two images would be aware of the conflicting but no less interesting relationship between resistance and power established between them.