Asrar

Future Nobel Prize Laureate.

[Mihir Prabhudesai @mihirp98]

🚨 The era of infinite internet data is ending, So we ask:

👉 What’s the right generative modelling objective when data—not compute—is the bottleneck?

TL;DR:

▶️Compute-constrained? Train Autoregressive models

▶️Data-constrained? Train Diffusion models

Get ready for 🤿 1/n

[Asrar @_Asrarii]

@CVPR CVPR organizers: It’s not acceptable to push in-person attendees to Zoom after paying full price because the workshop room is too small. Please consider audience size when assigning rooms this isn’t what we paid for. :)

[Asrar @_Asrarii]

@overleaf 😭😭😭😭😭 we have deadlines guys hurry up

[Asrar @_Asrarii]

@wacv_official You mean Jan 2025 !?

[Costa Christopoulos @costhegreek]

Your thesis committee in the last year of your PhD:

[Asrar @_Asrarii]

@costhegreek 😂😂😂😂

[Asrar @_Asrarii]

We made it 🎓

[Carnegie Mellon University @CarnegieMellon]

The world awaits, Tartans ❤️

One week until we celebrate your accomplishments at commencement!

#CMUGrad

[Asrar @_Asrarii]

@LifeAfterMyPhD 😂😂😂

[Gerard Pons-Moll @GerardPonsMoll1]

GENERATING CONTINUAL HUMAN MOTION IN DIVERSE 3D SCENES @3DVconf '24

1) we canonicalise motion and train a model to converge to the origin (0) of the next waypoint coordinate frame, which makes it drift-free.
2) We drive motion based target keypoints in the 3D scene

[Asrar @_Asrarii]

@LifeAfterMyPhD My little boy wishes you well soon 🐱

[Asrar @_Asrarii]

@CSProfKGD When will we receive tickets?

[Samsung Research @samsungresearch]

[Tech Blog] #CVPR2023 series ①

Introducing "SPIn-NeRF: Multiview Segmentation and Perceptual Inpainting with NeRFs" accepted at CVPR 2023.

#NeRF #Neural_Rendering #Inpainting #3D_Segmentation #Novel_View_Synthesis #Multiview_Segmentation #ComputerVision
https://t.co/5s9pGut0FD

[MIT Mathematics @MITMath]

Prof. Gil Strang's last lecture is ... Monday! Watch it via live stream: https://t.co/crToqS9gKq

[Chelsea Finn @chelseabfinn]

I gave a talk in the @CMU_Robotics seminar on
* robot generalization via broader data
* generalizing beyond the train env through adaptation & invariances
* on-going work on dexterous tasks

Thank you @deepakpathak for hosting & others for great discussions!
https://t.co/eCs59lgqDY