3/3 It's still early for this task; Help us make these panoptic segmentation predictions perfect and realize the downstream impact: https://t.co/miz7u0twtN

2/3 The multicam + video data, temporal continuity of a slowly moving viewpoint, close collaboration with data sourcing and labeling, and the infinity-sized dataset of unlabeled clips dramatically expands creative modeling opportunities on the neural net side

1/3 Some panoptic segmentation eye candy 🌈🤩 from a new project we are bringing up. These are too raw to run in the car, but feed into auto labelers. Collaboration of data labeling a large (100K+), clean, diverse, multicam+video dataset and engineers who train the models

Programmed alongside GitHub Copilot (https://t.co/Bpl111vX78) for a while now and have to say I am very impressed and find myself tab-completing a lot of code. E.g. following chunk was a tab complete (except I manually fixed a bug of > into >=). Overall excellent and eerie.

Google Maps figuring out how to shave 1 minute off your trip with only 10 extra steps and twists and turns be like 🧐😳😱😱🤯🤩🤩🤩🥳

Haha, wanted to start a dev server so I was going to `make run`, but misspelled it as `make fun`, and then decided this is much better and changed the Makefile. Anyway, happy thanksgiving everyone! :) 🙏

Netflix's 'Arcane' (from Riot Games, of League of Legends fame) is refreshingly and unepectedly beautiful. (am on ep4) https://t.co/M806TfH4ft

cc @paperswithcode love your work a lot, this may be my biggest consistent struggle. Don't want to see the highest scoring, but the one you actually want to use. Often a struggle to identify. Closest proxy I know is the 2nd best approach everyone seems to compare themselves to.