We are tremendously excited to announce that as of this afternoon, the GDELT Visual Knowledge Graph (VGKG) has reached 46 million images processed through Google's Cloud Vision API deep learning neural network algorithms, making it what we believe is the largest open dataset of deep learning processing of images. Each day GDELT feeds more than half a million images from online news articles published in every country of the world in 65 languages through Google's deep learning algorithms to annotate them with the objects and activities they depict, recognizable text, geographic location, logos, and even the emotions of each human face present.
You can download the data today!
Stay tuned for several incredibly exciting announcements relating to this dataset! In the meantime, here are some links to more information about the dataset:
- Forbes: Using Google's Deep Learning AI To Geolocate Global News Imagery (Technical Details)
- Visual Global Knowledge Graph (VGKG) February 2016 Snapshot Dataset
- VGKG 1.0 Arabic, Persian, Thai, Chinese And Script OCR
- VGKG 1.0 Now Processing All Monitored News Imagery
- GDELT Visual Knowledge Graph (VGKG) V1.0 Available
- Washington Post: What Does Artificial Intelligence See When It Watches Political Ads? (Technical Details / Image Frames)
- Forbes: Mapping World Happiness And Conflict Through Global News And Image Mining
- Tracking The Red Cross Logo
- 72 Hours Of Fire Imagery: Saudi Protests & Yemen Bombings
- Google BigQuery + Visual GKG: Sample Queries
- Announcing The New GDELT Visual Global Knowledge Graph (VGKG)
- 48 Hours Of Pollution And Littering Around The World Through Deep Learning
- Deep Learning Triaging For Disaster Response
- Deep Learning Image Identification To Counter Poaching
- Image-Based Georeferencing: Recognizing Locations From Images
- Triaging Disaster Imagery: Cyclone Pam & Vanuatu
- Exploring The Threads Of Visual Narrative
- Emotions From Images: Sentiment Mining The World's News Imagery
- A Vision Of The Future: GDELT + Google Cloud Vision API