Shing Ho J. Lin

MPhil @ Tsinghua Automation

[Shing Ho J. Lin @jojojo424242]

@wenhaocha1 Genius

[Shing Ho J. Lin @jojojo424242]

@coralli04102979 perfect name 🤣

[Shing Ho J. Lin @jojojo424242]

@X0wcol1 @PeterHndrsn And I’m not saying not being SoTA should be rejected. There’s a huge gap between “sub-SoTA” and “left behind for 5 yrs”. Besides, this work didn’t even reproduce the 2021 model correctly and the performance was degraded. So declaring itself as SoTA doesn’t stand even in 2021.

[Shing Ho J. Lin @jojojo424242]

@X0wcol1 @PeterHndrsn Well the problem is that the task is far from niche (I can’t go anymore specific). This has nothing to do with compute, as recent methods takes similar amount of compute.

[The Nobel Prize @NobelPrize]

WATCH LIVE: Join us for the announcement of the 2026 Nobel Prize in Physics.

Hear the breaking news first.

Where are you watching from?

#NobelPrize https://t.co/sgLBdJRy2P

[Shing Ho J. Lin @jojojo424242]

@NobelPrize My guess is Dennis Lo ✌️

[The Nobel Prize @NobelPrize]

WATCH LIVE: Join us for the announcement of the 2026 Nobel Prize in Physiology or Medicine.

Hear the breaking news first.

Where are you watching from?

#NobelPrize https://t.co/epIMr3uiSH

[Ilya Fradlin @ifradlin]

(1/8)
We are excited to share our new model ARROW 🏹 !

ARROW brings D4RT-style decoding to broader input types, including multi-view videos and unordered collections of images.

📜arXiv: https://t.co/KJKeQdmo0R
🌐Project page: https://t.co/3T3kxMnTHH

[Shing Ho J. Lin @jojojo424242]

@ifradlin Great work! Thanks for citing us 🤓

[Einsia @EinsiaAI]

The next test for AI agents is not just whether they can write code.
It is whether they can look, understand, code, inspect—and fix.
Today we’re releasing PPTBench, a benchmark for Visual Coding through scientific diagram slides reconstruction.

500 tasks. 36 configurations. 18,000 reconstructions.
64.03% failed the semantic check.

Only 2.57% passed all gates with no recorded defect.

Visual coding is still wide open.

🤓 Read agent sessions on AgentGit: https://t.co/XQppVwg6vV

Full breakdown 👇
Project: https://t.co/8nkiEZKFq9
Paper: https://t.co/PC757AHyMG
Github: https://t.co/jnsSLsYfdl

[Shing Ho J. Lin @jojojo424242]

@sarahwiegreffe Two identical paper (title and abstract) with different numbers appeared in my pool 😆

[Shing Ho J. Lin @jojojo424242]

@PeterLauLukCh @iclr_conf Got it. Even if the guideline didn’t specify, it’s still intolerable 😆

[Shing Ho J. Lin @jojojo424242]

@fuxinli2 During that period, I would suggest strictly desk rejecting irresponsible reviewers (the revealed number of desk rejected article due to irresponsible review is far from strict) as I’ve encountered one exactly during CVPR25. Now there’s really nothing we can do.

[Letian Wang @Letian_Wang_6]

Can video generation models do for vision what LLMs did for language?

Introducing GenCeption from @GoogleDeepMind: one feed-forward video model for various vision tasks — SOTA, data-efficient, and emerging behaviors (ECCV 2026)

🌐 https://t.co/3rwVgTolZz

(1/8)

[Shing Ho J. Lin @jojojo424242]

@n0tr0uble @HuggingPapers almost 3yrs ago👀

[Minsik Jeon @Minsik_Je0n]

[1/N] Introducing Point4D: dense 4D reconstruction across hundreds of frames from a monocular video. Unlike existing feed-forward 4D methods that operate over short windows, Point4D keeps tracking in 3D, even across occlusions and out-of-frame motion.

[Shing Ho J. Lin @jojojo424242]

@Minsik_Je0n Really nice to see the advancement of D4RT getting open-sourced! Especially when it doesn’t 😆