CoTracker3 (fastforks)

CoTracker3 (Karaev et al., Meta AI 2024): tracking any point with a transformer that tracks all points jointly. A CNN encodes every frame at 384 x 512 into 128-channel features at 1/4 resolution (a 4-level pyramid); each track correlates the 7 x 7 neighbourhood of its current position with the 7 x 7 support of its query on every level (an MLP encodes the 4D correlation), and a transformer with attention over time and, through 64 virtual tracks, across tracks updates positions, visibility and confidence in 6 iterations. Offline: the whole video at once; online: sliding windows of 16 frames, 8 new frames per step.

This repo is the model card only: it holds no weights. fastforks downloads the authors' weights from the links below and checks them against the SHA-256 in the model card. Code: https://github.com/kleinicke/fastforks. Kernels for the kernels backend: https://huggingface.co/fastforks/kernels. Original code: https://github.com/facebookresearch/co-tracker at 82e02e802975.

Author weight repos: https://huggingface.co/facebook/cotracker3.

Task: tracking. Tier: experimental. Outputs: tracks, visibility. Backends: portable, kernels.

Input: a video (its frames in order) and query points (t, x, y). Output: the position of every query in every frame and the probability that it is visible there.

Variants

Weights are the authors' files (not hosted here). SHA-256 shortened to 12 hex digits; the full hashes are in card.toml.

variant weights (author link) SHA-256 weights licence commercial use
cotracker3-offline@1 scaled_offline.pth 2670d4562ed6 CC-BY-NC-4.0 no
cotracker3-online@1 scaled_online.pth 205d34789f19 CC-BY-NC-4.0 no

Licence

Original-code strings, recorded as they are (research-use package; check them before any other use). Code: CC-BY-NC-4.0. Licence files: LICENSE.md.

Training data: Upstream paper: trained on synthetic Kubric MOVi-F sequences, then fine-tuned by pseudo-labelling about 15k real internet videos with other trackers (not released). No dataset entry here yet..

Licence evidence: https://github.com/facebookresearch/co-tracker/blob/82e02e8029753ad4ef13cf06be7f4fc5facdda4d/LICENSE.md (CC BY-NC 4.0); https://github.com/facebookresearch/co-tracker/blob/82e02e8029753ad4ef13cf06be7f4fc5facdda4d/README.md#license; https://huggingface.co/facebook/cotracker3/blob/bf55ea50d4390e1820a267f131cd6587240fb2c5/README.md (license: cc-by-nc-4.0)

Input size

any H x W >= 32, padded to a multiple of 32 Tested up to (portable, fp16, RTX 5060 Ti 16 GB): cotracker3-offline 3000x4000 (5.5 GiB peak allocated, 5.9 reserved); cotracker3-online 3000x4000 (5.5 GiB peak allocated, 5.9 reserved)

Speed and memory

Tracking, synthetic, 480x854

model time (ms): original portable kernels kernels speed-up: vs original vs portable peak memory (MiB, allocated / reserved): original portable kernels
cotracker3-offline 244 162 154 1.6x 1.1x 2221 / 2480 755 / 882 799 / 916
cotracker3-online 328 184 176 1.9x 1.0x 1659 / 1886 755 / 906 798 / 922

Startup and first call

model load (s): original portable kernels first call (s): original portable kernels
cotracker3-offline 0.65 0.28 0.19 0.87 0.42 0.52
cotracker3-online 0.54 0.65 0.18 0.54 0.39 0.54

torch.compile and TensorRT

model input time: portable compile TensorRT fp16 kernels preparation: compile (cold first call) ONNX + engine build peak MiB (alloc / reserved): portable compile kernels TensorRT (engine + torch)
cotracker3-offline 480x854 162 ms 135 ms 25.9 ms 154 ms 36.5 s 157 s 755 / 882 755 / 902 799 / 916 519 / 586
cotracker3-online 480x854 184 ms 159 ms planned for 0.2.1 176 ms 37.7 s – 755 / 906 701 / 846 798 / 922 –

Fixed-size cells: time (ms) / peak allocated / peak reserved memory (MiB).

Hardware, protocol and caveats: benchmarks (RTX 5060 Ti 16 GB, sm_120). Original = the original code named above at the precision of portable (fp16, or (fp32) where fp16 is unsupported or not finite); original fp32 = its default fp32 run. Peak memory is the torch allocator's.

Accuracy

On public datasets

TAP-Vid protocol: frames resized to 256 x 256 (PIL Lanczos) before the model (AllTracker: the original frames, resized by the model), one query per track at its first visible point, tracked online (TAPNext, Track-On) or over the whole clip (TAPIR, AllTracker, LocoTrack); AJ, delta_avg (< 1, 2, 4, 8, 16 px at 256 x 256) and occlusion accuracy averaged over the videos. Original = the original code named under Details, same checkpoint and protocol.

variant dataset original fp32 original fp16 portable kernels note
cotracker3-offline TAP-Vid DAVIS (query first), every 3rd video, AJ / delta_avg / OA % 65.02 / 78.75 / 88.24 65.08 / 78.78 / 88.27 65.07 / 78.75 / 88.26 65.06 / 78.72 / 88.27
cotracker3-online TAP-Vid DAVIS (query first), every 3rd video, AJ / delta_avg / OA % 70.34 / 80.63 / 94.28 70.28 / 80.62 / 94.26 70.32 / 80.62 / 94.26 70.25 / 80.63 / 94.22

Measured on NVIDIA GeForce RTX 5060 Ti, torch 2.14.1+cu130, fastforks f104019b, 2026-11-14.

Parity with the original code

Every variant is checked against reference outputs of the original code at its pinned commit. CPU, float32: max |diff| <= 6e-05 vs the original. GPU, float16 (the default), portable and kernels: mean |diff| <= 0.0027 and p99 <= 0.014 (absolute error of the output tracks).

Limitations

  • Frames are resized to 384 x 512 (the training size and that of upstream's predictor), whatever their size and aspect ratio.
  • cotracker3-offline holds the features of every frame (about 4 MB per frame in float16); cotracker3-online processes 16-frame windows and only keeps the tracks.
  • Visibility is upstream's score, rescaled so that 0.5 is upstream's threshold: the offline predictor's visibility > 0.9, the online predictor's visibility x confidence > 0.6.
  • One video per call; all queries of a call share the video.

Usage

from fastforks import tracking

# non-commercial weights: run with FASTFORKS_NONCOMMERCIAL=1 in the environment
model = tracking.load("cotracker3-offline")  # mode="auto": the preferred backend from recorded measurements
result = model(frames, [[0, x, y]])  # frames: the video's images in order; queries (t, x, y)
result.tracks  # [T, Q, 2] float32: x, y in input pixels; result.visible [T, Q] bool
result.backend  # what ran, e.g. Backend(mode='portable', device='cuda:0', dtype='float16')

Install with pip install fastforks (see https://github.com/kleinicke/fastforks). Other variants: cotracker3-offline, cotracker3-online.

Kernels backend

tracking.load("cotracker3-offline", mode="kernels") runs our CUDA kernels (fvm_conv). It needs a prebuilt kernel for your GPU architecture (downloaded from https://huggingface.co/fastforks/kernels on first use) and a recent NVIDIA driver; if there is none it raises BackendUnavailableError with the reason and never falls back silently. mode="auto" picks kernels when it can run and logs the choice.

Downloads last month

-

Downloads are not tracked for this model. How to track
Inference Providers NEW
This model isn't deployed by any Inference Provider. πŸ™‹ Ask for provider support

Model tree for fastforks/cotracker

Finetuned
(3)
this model

Collection including fastforks/cotracker