←
Adam W. Harley284 posts
Adam W. Harley avatar

Adam W. Harley

@AdamWHarley

Research Scientist at Meta. I did a postdoc at Stanford and PhD at CMU. I work on computer vision and machine learning.

◷ Joined August 2020▧ 95 Photos and videos
Adam W. Harley avatar
Adam W. Harley@AdamWHarley·Jun 19

Introducing VOTSp2026 — a new point tracking challenge! 50 hand-picked videos including animals, robots, cells, egocentric capture, and more, with clean annotations and standardized evaluation. Submit your tracker and compete: votchallenge.net/vots2026/

0:10
Ayush Jain avatar
⟳ Adam W. Harley retweeted
Ayush Jain@ayushjain1144·Oct 20

We released the code and model weights here: github.com/ayushjain1144/tra…. Have fun! 🙂

0:06

1/ We introduce TrackEverything: a 3D point tracker that tracks all points across all frames of long videos (1000+ frames). Our key idea is to tie computation to unique 3D scene content rather than redundant 2D pixels in a video. trackeverything.github.io

0:47
Adam W. Harley avatar
Adam W. Harley@AdamWHarley·Oct 14

Our team is hiring PhD interns now for 2027. Ping me if you are interested and qualified, especially if you work on similar problems as me.

Project Aria @Meta avatar
⟳ Adam W. Harley retweeted
Project Aria @Meta@meta_aria·Sep 25

Check out TrackEverything from CMU Robotics. 💡 TrackEverything is a Long-term 3D point tracking method (1000+ frames), using depth and camera pose to see the world in a stable coordinate system. TrackEverything is especially effective for egocentric capture setups like Project Aria glasses. Amazing work by @ayushjain1144 while interning with @AdamWHarley at Meta. 🚀

1/ We introduce TrackEverything: a 3D point tracker that tracks all points across all frames of long videos (1000+ frames). Our key idea is to tie computation to unique 3D scene content rather than redundant 2D pixels in a video. trackeverything.github.io

0:47
Ayush Jain avatar
⟳ Adam W. Harley retweeted
Ayush Jain@ayushjain1144·Sep 25

1/ We introduce TrackEverything: a 3D point tracker that tracks all points across all frames of long videos (1000+ frames). Our key idea is to tie computation to unique 3D scene content rather than redundant 2D pixels in a video. trackeverything.github.io

0:47
Adam W. Harley avatar
⟳ Adam W. Harley retweeted
Adam W. Harley@AdamWHarley·Jun 22

The deadline for participating in the VOTSp2026 point tracking is this Sunday, June 28! Preparing a submission should take a couple hours max (especially with a coding agent involved). Give it a shot! The results of the challenge will be presented at ECCV 2026.

Introducing VOTSp2026 — a new point tracking challenge! 50 hand-picked videos including animals, robots, cells, egocentric capture, and more, with clean annotations and standardized evaluation. Submit your tracker and compete: votchallenge.net/vots2026/

0:10
Adam W. Harley avatar
Adam W. Harley@AdamWHarley·May 21

Very fun: Our paper for Insect Science is one of the journal's most-viewed articles of 2025! Yay for point tracking + vibrational signal analysis (with @JPolajnar)! BTW the underlying point tracker here is basically "PIPs+++" -- a simple method tuned well can do a lot!

Post attachment
1:00
Project Aria @Meta avatar
⟳ Adam W. Harley retweeted
Project Aria @Meta@meta_aria·May 14

The Project Aria team is excited to be part of the Third Joint Egocentric Vision (EgoVis) Workshop at #cvpr2026! 👓✨ Don’t miss the latest Project Aria updates as we dive into the future of egocentric perception. Check out the full program of invited speakers below! 📅 June 3, 2026 📍 Colorado Convention Center | Room 704/706 🔗 egovis.github.io/cvpr26/ ⚡️@liuziwei7 , @mapo1 , @doughty_hazel

Post attachmentPost attachment
Tali Dekel avatar
⟳ Adam W. Harley retweeted
Tali Dekel@talidekel·Mar 30

Dynamic 3D from monocular video 🎥 Static pre-scan + video = richer reconstruction ✨ As image-to-3D improves, static pre-scans become increasingly practical! Kudos to @tnarek99 and team! drops-dynamics.github.io

Excited to present DRoPS! A novel method for dynamic 3D reconstruction from monocular videos, accompanied by a static-prescan of the dynamic object. 🌐 Project: drops-dynamics.github.io 📄 Paper: arxiv.org/abs/2603.24770 🧵 [1/7]

0:26
Xindi Wu avatar
⟳ Adam W. Harley retweeted
Xindi Wu@cindy_x_wu·Jan 20

New #nvidia Paper We introduce Motive, a motion-centric, gradient-based data attribution method that traces which training videos help or hurt video generation. By isolating temporal dynamics from static appearance, Motive identifies which training videos shape motion in video generation. 🔗 research.nvidia.com/labs/sil… 1/10

1:15
Yawar Siddiqui avatar
⟳ Adam W. Harley retweeted
Yawar Siddiqui@yawarnihal·Jan 19

Introducing ShapeR, a method for robust conditional 3D shape generation from casually captured sequences. ShapeR leverages a rectified flow transformer conditioned on per-object multimodal data to turn casual image sequences into full metric scene reconstructions. Project Page: facebookresearch.github.io/S… Paper: arxiv.org/abs/2601.11514 Links to code and huggingface below ⬇️

0:59
Dima Damen avatar
⟳ Adam W. Harley retweeted
Dima Damen@dimadamen·31 Oct 2025

📢 New Paper PointSt3R: Point Tracking through 3D Grounded Correspondence Can point tracking be re-formulated as pairwise frame correspondence solely? We fine-tuning MASt3R with dynamic correspondences and a visibility loss and achieve competitive point tracking results 1/3

0:25
Adam W. Harley avatar
Adam W. Harley@AdamWHarley·12 Jul 2025

Yup! But I wouldn't say "single shot" -- AllTracker, inheriting from CoTracker and PIPs and RAFT, is an iterative approach. In practice, we iterate 4x to get the final answer (and this sums to 11 FPS at 576x1024).

AllTracker. Hi-res, dense point tracking across long video ranges in a single shot. Tracks every pixel. Fast, lightweight (16M params). Sparse tracking is obsolete. alltracker.github.io/

0:33