OpenAI trained a neural network that they call DALL·E with a dataset of text and image pairs. So now the neural network can take text input and output random combinations of descriptors and objects, like a purse in the style of Rubik’s cube or a teapot imitating Pikachu. Tags: images, neural network, OpenAI, text
Generative models can seem like a magic box where you plug in observed data, turn some dials, and see what the computer spits out. SpaceSheet is a simple spreadsheet interface to explore and experiment for a clearer view of the spaces between. Even if you’re not into this research area, it’s fun to click and drag things around to see what happens. Tags: generative models, images, latent space
Triangulate, a fun tool made by Michael Freeman, lets you upload a picture and it randomly assigns points to output something that looks pixelated but with triangles. Give it a try. Tags: Delaunay, images
Traditional detection algorithms use infrared heat as the main signal of a wildfire. The Firelight Detection Algorithm uses visible light instead, detecting a fire possibly a day earlier. FILDA uses the visible light of fire, detecting high resolution images of fires. Using VIIRS technology, images of fires at night can be captured using infrared and visible light information. FILDA can also detect 90% more pixels than previous methods, and can...
NeuralTalk2 uses neural networks to caption images quickly. To demonstrate, the video below shows a webcam feed that continuously updates with new image captions based on what the computer sees. It's not perfect of course, but the performance is impressive. Tags: classification, images, neural networks