Computer Vision news and analysis on Web Pulse: 4292 curated articles tracking the latest Computer Vision developments, tools, and research, updated continuously from vetted sources.
A developer built a small Pillow and OpenCV script for evaluating outdoor portrait generators, finding that whole-frame Laplacian variance misleads because blurred backgrounds dominate the image. On a calibration dayligh…
A developer built TrailShield, a phone-only web app that uses Google's open-weight Gemma model (gemma-4-26b-a4b-it) to give short, cautious notes on trail photos and then prompts users to put their phone away. The app st…
A developer built SideQuest, a Next.js app that uses Google's Gemma 4 26B model to generate short outdoor missions and then prompts users to put their phone away, with a Gemini 3.7 Flash verifier checking a single return…
A developer built Seekr, an open-source, map-based treasure hunt app for real-world events in which organizers pin riddles and hidden locations and teams of up to four solve clues, walk to each spot, and verify arrival w…
Alibaba's Qwen team released Qwen-Image-2.1, a roughly 20-billion-parameter multimodal diffusion model that renders legible, correctly spelled text inside generated images across Chinese, English, Japanese and Korean. Th…
A developer built Touch Grass, an AI-powered outdoor scavenger hunt that uses a player's location and nearby OpenStreetMap geographic context to generate five objects to find, verified through a YOLO detection pipeline. …
Tamoghna Chattopadhyay and colleagues at the University of Southern California published MEMOIR-VLM on October 1st, a research model that classifies Alzheimer's disease from whatever combination of MRI, diffusion imaging…
Developer Vytas Gavelis built Hue Hunt, an open-source daily colour-hunting game for friend groups in which players photograph objects matching a daily colour and bonus prompt, with photos hidden until an evening reveal.…
A developer built Verdantra, an open-source AI-powered botanical field journal that uses photo analysis to guide outdoor observation rather than replace it. The project pairs a React frontend with a FastAPI backend and a…
A developer built OUTCURIO, an AI-powered "real-world curiosity companion" that uses a photo of a user's surroundings and a vision-language model to generate offline activities meant to reduce screen time. The project pa…
Stanford PhD student Ziga Kovacic and collaborators introduced 4DCodeBench, a 200-scene benchmark testing whether coding agents can turn video of a physical event into executable graphics code reproducing geometry, appea…
A developer built WatermarkLift, a free browser-local tool that removes the visible bottom-right Gemini star overlay by inverting the alpha-blend equation using the pinned MIT-licensed @pilio/gemini-watermark-remover@1.0…
A developer built RapSong, a SaaS tool that turns two static photos into a Hotel Lobby-style rap duet video by combining lyric writing, beat production, lip sync and character identity preservation in a single pipeline. …
Team ARKA built FitGuru, an Android workout coach that runs MediaPipe pose estimation entirely on-device, using joint-angle state machines for rep counting and form cues while only sending structured JSON (angles, reps, …
A developer built a Kaggle benchmark testing 11 frontier vision-language models on zero-shot OCR of ISO metric tire size codes from 51 curved tire sidewall images across 22 dimensions. Google Gemini 3.8 Flash and Anthrop…
Researchers in Prof. Michal Irani's lab at the Weizmann Institute of Science developed Brain-IT, an AI model that reconstructs viewed images from fMRI brain activity and needs only one hour of scans to learn to read a ne…
A developer built WildHunt AI, an open-source, privacy-first scavenger-hunt progressive web app that uses a local open-weight vision model to verify photos users take of real-world targets. The app issues missions across…
A developer built NaturDex, an offline, on-device AI nature identification app that turns a recycled Android phone into a Pokédex-style field device for school trips. The app runs an image classification model locally, a…
A developer built Touch Grass, a single-user app that assigns one daily outdoor task and verifies completion with a live photo checked by a local vision-language model. The app runs Qwen2.5-VL 3B through Ollama on an RTX…
A developer built Proof of Grass, a daily outdoor bingo card web app that uses CLIP ViT-B/32 quantized to 8-bit and run in-browser via transformers.js to verify photos of outdoor items. Testing against 131 Wikimedia Comm…