PRCV Labs

Perception Research in Computer Vision. An open source org exploring and advancing large scale CV systems, particularly in tracking and action recognition.

[Everett @retttx]

Congrats @katylevinson @sidesolve! https://t.co/DjenCIxVMj

> The relator, Sidesolve, Inc., will receive a $1 million share as part of the settlement.

[Kenneth Cassel @KennethCassel]

I think robotics will be peak vc hype in 2-3 years

[Dan Grover @DanGrover]

People are dumping on GPT wrappers, but tons of robot/machine vision companies successfully charging $5K+/deployment for OpenCV wrappers, and not even good ones.

[Marc @mrousavy]

Built with VisionCamera. All the drawing and blurring is happening with JavaScript code, and it runs in realtime. No post-processing. This is going to change the entire mobile camera game. Get ready.

[PRCV Labs @prcvlabs]

LangChain including 'experimental code' but the whole library is experimental code

[LangChain @LangChain]

You can now import BabyAGI, AutoGPT, and the generative agents memory from `langchain.experimental`!

We started this module to include more experimental code as we think about how best to include in the core library

What should we include next?

[PRCV Labs @prcvlabs]

Using LLMs effectively for work boils down to understanding 1 important limitation:
https://t.co/X9yGZkkCtk

[Paul Graham @paulg]

https://t.co/SH2bntXGfN

[Rowan Cheung @rowancheung]

Is this the future of the metaverse?

This new generative AI tool creates text-to-space.

You can now create interactive virtual experiences through simple prompts.

[evan sc / 柳爱华 @stitesclay]

After 7 hours - Why does it keep responding "As an AI assistant I am a large language model and as such I cannot ... xyz"

I didn't put that anywhere in the prompt where is it getting that from??

[Strange Loop Labs @strangeloopai]

We're launching Guru Sports, an AI-powered dev toolkit for providing elite sports analysis on mobile video.

Nearly half of the 2023 NFL Draft's 1st-round are already using Guru Sports to train for next week's #NFLCombine with @les7spellman. Learn more👇🏾

https://t.co/7Hd9s5biaF

[Mark Tenenholtz @marktenenholtz]

6. Reuse old Kaggle solutions

Kaggle solutions are like a database for battle-tested solutions.

If you have a similar enough problem to an old competition, those solution writeups are an absolute goldmine.

My first stop on any problem for "literature review" is Kaggle.

[Luma @LumaLabsAI]

Luka Dončić is back at it with Nerf, Luma and the amazing @CorridorDigital crew!
https://t.co/l6gw34Nny2

[mok @mokshith]

1/ Today we’re excited to unveil Sieve (@sievedata), a cloud native platform for processing, searching, and running all sorts of AI models on video.

https://t.co/VJe3lgBJuX

[Dmytro Mishkin 🇺🇦 @ducha_aiki]

Floorplan-Aware Camera Poses Refinement

Anna Sokolova, Filipp Nikitin, Anna Vorontsova, Anton Konushin

tl;dr:
If you have a map or floorplan, use it for improving 3d reconstruction and camera trajectory

https://t.co/ruedEJgzhj

[Aperture Robotics @ApertureRobo]

Bifrost: cross-platform p2p communications engine in Go. Modular daemon & library for web3.0 connectivity and @libp2p peers. Supports RPC & NATS PubSub, Quic, and running in the web browser. https://t.co/lhgD4umY9a

[Jack Merullo @jack_merullo_]

Are language models (LMs) good models of the visual world? We show that without explicit grounding, LMs can directly use linear projections of image representations as soft prompts for vision-language (VL) tasks. This can be done without tuning the LM or image encoder!

[roon @tszzl]

seems like tesla has given up on “no deep maps” ideology. now they’re just mapping and NeRFing everything

[PRCV Labs @prcvlabs]

> computer vision is one of the most significant technological steps in the last ten years and has fundamentally changed the scale of innovation.
https://t.co/5Yk8XDjwyu