Singing Data @SingingData

@Variety Oh please, he was on several sketches every Saturday.

I got you covered Explore all CVPR papers here: Oral papers, Spotlight papers, and all papers categorized by domain paperswithcode.co/conferences/cv…

Chester @chesterzelaya

2 days ago

was going to look over the CVPR papers this year.. then i found out there are 5022 papers accepted

2 0 13 88K 2

Breaking News: For the first time ever, Mister Rogers’ Neighborhood has officially launched a YouTube channel. Full episodes. Classic Clips. 100% wholesome. 💗

All 15 @CVPR 2026 Paper Finalists can now be easily explored here paperswithcode.co/conferences/cv… Find GitHub links, @huggingface artifacts, and evals

#CVPR2026 @CVPR

3 days ago

Finalists 👏

1 5 48 49K 18

My cancer docs are celebrating all the new drugs coming into the armamentarium, but they say in 2-6 years there’s going to be a lag because do much research has been stopped nytimes.com/2026/06/05/wel… via @NYTimes

It has been a blast sharing the excitement, science, art, and community of CVPR with all of you. The Publicity Chairs are officially signing off. Seattle, you’re up next! 🌲 See you at #CVPR2027! 👋 @deblinaforAI @anfurnari @CSProfKGD @YVinker @_vztu

This was one of the standout AI papers of the week. (bookmark it) It tackles a question most self-improving AI agents ignore: is the agent actually discovering anything, or just remixing what it already knows? How can you tell whether the agent is doing real discovery or just confident retrieval? The authors give three clean buckets: - Retrieval is looking something up in a notebook you already have. - Search is combining tools you already own in new ways. - Discovery is inventing a new concept that wasn't in your toolkit before. The issue is that most agents stop at the first two. The math behind their definition (category theory plus a left Kan extension, if you care) is basically a bookkeeping trick to ask: could the old version of me have produced this result? If yes, it's not discovery. If no, something genuinely new showed up. They build a Builder/Breaker agent that studies protein mechanics. Over four rounds, the model's fit accuracy actually drops (R² goes from 0.48 to 0.68 to 0.54 to 0.41). At first glance, that looks like a failing agent. It isn't. The agent kept taking on harder proteins and rewriting its theory to cover them. Data grew almost 10x while the model code grew only 1.3x. A smaller theory covering a bigger world is exactly what good science looks like. Why does it matter? If you optimize for accuracy alone, your self-improving agent will just settle into easy benchmarks and stop. This paper offers a cleaner success signal and asks whether the agent is compressing more of the world into less code over time. Paper: arxiv.org/abs/2606.01444 Learn to build effective AI agents in our academy: academy.dair.ai

@RaminNasibov I mean, maybe it's waiting - lying latent until the species needs to start chewing raw meat again.

"We were going out stealing horses. That was what he said, standing at the door to the cabin where I was spending the summer with my father. I was fifteen. It was 1948 and one of the first days of July. Three years earlier the Germans had left, but I can’t remember that we talked about them any longer. At least my father did not. He never said anything about the war. Jon came often to our door, at all hours, wanting me to go out with him..."

Great paper on self-improving agents:

elvis @omarsar0

a day ago

This was one of the standout AI papers of the week. (bookmark it) It tackles a question most self-improving AI agents ignore: is the agent actually discovering anything, or just remixing what it already knows? How can you tell whether the agent is doing real discovery or just

27 68 334 38K 389

INSID3 segments objects across domains using ONLY ONE annotated example it works entirely without a segmentation decoder, task-specific fine-tuning, or external mask generators like SAM CVPR 2026 paper with enormous practical potential

@buccocapital and they've been working on speech recognition since at least 1991 when I interned there. 35 years.

I think this is my favorite paper this CVPR: Magician. before they explore in active view selection, they imagine how gaussians and occupancy map would look like and then compute a coverage metric based on that. during planning, they try 10 views like that in 10 steps in a tree search with pruning and get planning for free. they even have real-world experiments with a drone and a toy car. how are they not an award candidate, it blows my mind.

THE COMPUTER VISION TOOLKIT BUILT FOR ML ENGINEERS roboflow/supervision gives you all the reusable building blocks for cv pipelines object detection, tracking, annotation, dataset management so you stop rewriting the same utils for every project. model agnostic. plug in yolo, transformers, mmdetection, or inference and it just works. load datasets in coco, yolo, or pascal voc. split, merge, convert formats in a few lines. annotate with bounding boxes, masks, labels, and more fully customizable. real-world use cases people have shipped with it: football player tracking, vehicle speed estimation, traffic zone analysis, dwell time measurement. 40k stars. 6.6k projects using it in production.

Silence can help us most to recognize the voice of God, since it fosters attention and recollection. Freed from the noise of a thousand voices, we come to recognize that some voices deceive our desires, others buy us without nourishing us, and still others speak out of self-interest. In silence, we understand that ideologies pass away, while truth remains. vatican.va/content/leo-xi…

What's amazing to me is that, 10 years apart, there are the exact same clouds in the exact same positions. DC truly is a magical place

📢📢📢We introduce Efficient-SID⚡️: training-free single-image diffusion model that generates images by sampling directly from an input image's patch distribution. Our method enables megapixel generation in <1s and scales to gigapixel generation. We also enable stylization, editing, and other applications. The outputs are constrained to follow exactly the patch distribution of the input — something that is very difficult to do with large models! #CVPR2026 Highlight 🌐 haojunqiu.github.io/efficient-SID/ 📄 arxiv.org/abs/2606.04299 [1/6]

Mondays after terrible Fridays for the NASDAQ

The Dog 🐶 @TAtheDog

3 days ago

@OddStats How many of those down days happened on a friday and what happened monday, if you have those odds

3 0 16 400K 2

Remember when they said Zelensky desecrated the sanctity of the White House by not wearing a suit

I want to offer some unsolicited advice to computer vision researchers jumping into robotics. Don't focus too much on VLMs, VLAs etc. That's fine, but the real action is at the sensorimotor level. Most of the open problems in robotics are in manipulation, which is about hand-object interaction, and contacts and forces are central. Proprioception and tactile sensing are as important as vision. Don't get seduced by cherry-picked demos. You can't do robotics without doing robotics.