SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2306.05422 · ICCV · 2023

Tracking Everything Everywhere All at Once

Bharath Hariharan, Noah Snavely, Yen-Yu Chang, Aleksander Holynski, Qianqian Wang, Zhengqi Li, Ruojin Cai

arXiv · PDF · Open in the Atlas

Code that ran

8 of 12 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (7 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 7 have no contract verdict; 1 failure of our instrument; 4 no recorded run), measured 2026-11-22; repository links from the Papers with Code archive attributed under CC BY-SA, no affiliation; bodies inline only under a permissive licence as GitHub detects it. Not a reproduction of the paper's results.

"Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results. This record, dated, as JSON: /api/ran/2306.05422.json · badge: /api/ran/2306.05422.svg

Each repository's role is a label: a classifier's output recorded on the link, or, where no role is recorded, what the Papers with Code archive marked. The line under it names its basis. Neither says who wrote the code.

RepositoryRoleRan
qianqianwang68/omnimotion marked official in Papers with Code
Papers with Code archive
8 ran · 4 no recorded run
FunctionStatusWhere it lives
CouplingLayer Ran
ran on a synthesized input · constructed an object; no result was computed
qianqianwang68/omnimotion/networks/nvp_simplified.py
code served (permissive licence) · get_code("c65dad517130b6ae")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned an instance of its own class, CouplingLayer: the run constructed the object and did not call it, so no result was computed. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn as nn

torch.manual_seed(0)

# Build the map_st network
# It takes concatenated input [F, projected_y1] and outputs [s, t]
# F has dim 5, projected_y1 has dim 2 (since mask selects 2 elements from y which has dim 4)
# So input to map_st is dim 7, output should be dim 2 (for s and t, each dim 1)
map_st = nn.Sequential(
    nn.Linear(7, 16),
    nn.ReLU(),
    nn.Linear(16, 2)
)

# Build the projection network
# It takes masked elements of y (dim 2) and projects to some dimension
# Output needs to be concatenated with F (dim 5) to feed into map_st
# So projection output should be dim 2
projection = nn.Sequential(
    nn.Linear(2, 8),
    nn.ReLU(),
    nn.Linear(8, 2)
)

# Build the mask
# y has shape (batch, 4), mask should select which elements are kept
# Let's make it select the first 2 elements
mask = torch.tensor([[1.0, 1.0, 0.0, 0.0]])

ARGS = {
    'map_st': map_st,
    'projection': projection,
    'mask': mask
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type CouplingLayer — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a CouplingLayer output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
FixedPositionalEncoding Ran
ran on a synthesized input · constructed an object; no result was computed
qianqianwang68/omnimotion/networks/nvp_simplified.py
code served (permissive licence) · get_code("1a6d87dec0694745")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned an instance of its own class, FixedPositionalEncoding: the run constructed the object and did not call it, so no result was computed. The image it ran in was not recorded. The code mentions cuda, and the harness aliases cuda to the CPU for every sample; whether this run needed that is not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import numpy as np
from torch import nn

torch.manual_seed(0)

ARGS = {
    'input_dims': 4,
    'frequency': 3,
    'proj_dims': 5
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type FixedPositionalEncoding — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a FixedPositionalEncoding output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
GaborLayer Ran
ran on a synthesized input · constructed an object; no result was computed
qianqianwang68/omnimotion/networks/nvp_simplified.py
code served (permissive licence) · get_code("081200769972f762")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned an instance of its own class, GaborLayer: the run constructed the object and did not call it, so no result was computed. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import numpy as np
from torch import nn

torch.manual_seed(0)

ARGS = {
    "input_dims": 4,
    "proj_dims": 3,
    "alpha": 1.0,
    "beta": 1.0,
    "weight_scale": 128
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type GaborLayer — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a GaborLayer output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
GaussianRandomFourierFeatures Ran
ran on a synthesized input · constructed an object; no result was computed
qianqianwang68/omnimotion/networks/nvp_simplified.py
code served (permissive licence) · get_code("0423139d18b88c8e")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned an instance of its own class, GaussianRandomFourierFeatures: the run constructed the object and did not call it, so no result was computed. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np
import torch
from torch import nn

torch.manual_seed(0)

# GaussianRandomFourierFeatures takes:
# - input_dims: int (dimension of input features)
# - proj_dims: int (dimension of projection)
# - gamma: float (default 1.0)

input_dims = 4
proj_dims = 8
gamma = 1.0

ARGS = {
    'input_dims': input_dims,
    'proj_dims': proj_dims,
    'gamma': gamma,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type GaussianRandomFourierFeatures — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a GaussianRandomFourierFeatures output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
IdentityProjection Ran
ran on a synthesized input · constructed an object; no result was computed
qianqianwang68/omnimotion/networks/nvp_simplified.py
code served (permissive licence) · get_code("03795e2cd57066cf")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned an instance of its own class, IdentityProjection: the run constructed the object and did not call it, so no result was computed. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn as nn

torch.manual_seed(0)

ARGS = {
    "input_dims": 4
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type IdentityProjection — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a IdentityProjection output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
MLP Ran
ran on a synthesized input · constructed an object; no result was computed
qianqianwang68/omnimotion/networks/nvp_simplified.py
code served (permissive licence) · get_code("629af934004c8e21")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned an instance of its own class, MLP: the run constructed the object and did not call it, so no result was computed. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn as nn

torch.manual_seed(0)

ARGS = {
    'c_in': 4,
    'c_out': 3,
    'c_hiddens': [8, 6],
    'act': nn.LeakyReLU,
    'bn': nn.BatchNorm1d
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type MLP — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a MLP output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
ProjectionLayer Ran
ran on a synthesized input · constructed an object; no result was computed
qianqianwang68/omnimotion/networks/nvp_simplified.py
code served (permissive licence) · get_code("9c0f2bb2dadff53c")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned an instance of its own class, ProjectionLayer: the run constructed the object and did not call it, so no result was computed. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn as nn

torch.manual_seed(0)

ARGS = {
    "input_dims": 4,
    "proj_dims": 3,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ProjectionLayer — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a ProjectionLayer output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
get_projection_layer Ran
ran; our contract draft was wrong, not the code
qianqianwang68/omnimotion/networks/nvp_simplified.py
code served (permissive licence) · get_code("b482fa381fe588ea")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned GaborLayer. The image it ran in was not recorded. The code mentions cuda, and the harness aliases cuda to the CPU for every sample; whether this run needed that is not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import numpy as np
from torch import nn

torch.manual_seed(0)

ARGS = {
    'type': 'gabor'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type GaborLayer — a GaborLayer output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a GaborLayer output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
BaseProjectionLayer Not yet run
unverified (harvested, no recorded run)
qianqianwang68/omnimotion/networks/nvp_simplified.py
code served (permissive licence) · get_code("07dee3b503d83af9")
How it ran

no invocation recorded.

NVPSimplified Not yet run
unverified (harvested, no recorded run)
qianqianwang68/omnimotion/networks/nvp_simplified.py
code served (permissive licence) · get_code("474eb246a3798427")
How it ran

no invocation recorded.

apply_homography Not yet run
unverified (harvested, no recorded run)
qianqianwang68/omnimotion/networks/nvp_simplified.py
code served (permissive licence) · get_code("c3f7178247f7c6f6")
How it ran

no invocation recorded.

apply_homography_xy1 Not yet run
unverified (harvested, no recorded run)
qianqianwang68/omnimotion/networks/nvp_simplified.py
code served (permissive licence) · get_code("af4c411e0529b9c9")
How it ran

no invocation recorded.

Also present in other papers' code

Of 12 samples lifted from this paper's linked repositories, 12 carry a function identity and 0 do not. Of the 12 distinct functions with an identity, 3 are also present in other papers whose code we hold (3 with at least one paper dated earlier, 0 the same month, 0 later) and 9 are only here among the 52,977 papers whose code Syntology holds (as of 2026-09-26).

Only here means: among the 52,977 papers whose code Syntology holds (51,887 with lifted samples, 1,090 through their assemblies only), as of 2026-09-26, no other paper's lifted samples or current assembly bricks contain an identical function body after normalisation.

FunctionAlso present inEarliest-datedOthers named
BaseProjectionLayer
brick of this paper's assembly: model
1 other paper
1 dated earlier
2203.16529 (2022-03)
CaDeX: Learning Canonical Deformation Coordinate Space for Dynamic Surface Representation via Neural Homeomorphism
IdentityProjection
brick of this paper's assembly: model
1 other paper
1 dated earlier
2203.16529 (2022-03)
CaDeX: Learning Canonical Deformation Coordinate Space for Dynamic Surface Representation via Neural Homeomorphism
ProjectionLayer
brick of this paper's assembly: model
1 other paper
1 dated earlier
2203.16529 (2022-03)
CaDeX: Learning Canonical Deformation Coordinate Space for Dynamic Surface Representation via Neural Homeomorphism

Presence, not direction: an identical function body in two papers' repositories does not say who wrote it, which came from which, or that either paper's method rests on it. Matching is exact after normalisation, so a renamed or edited function is not matched. Dates are when papers appeared, not when code was written. Only .py files the harvester lifted are covered; notebooks are not read. A new-style arXiv id YYMM.NNNN dates a paper to the month 20YY-MM; otherwise Paper.year dates it to the year; otherwise it is undated. Dates are when the paper appeared, not when the code was written. Read 2026-11-22.

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

Figure 1: We present a new method for estimating full-length motion trajectories for every pixel in every frame of a video, as illustrated by the motion paths shown above. For clarity, we only show sparse trajectories for foreground objects, though our method computes motion for all pixels. Our method yields accurate, coherent long-range motion even for fast-moving objects, and robustly tracks through occlusions as shown in the dog and swing examples. For context, in the second row we depict the moving object at different moments in time.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2306.05422")
get_code_for_paper("2306.05422")
have("2306.05422")

The run record, dated, one paper per request, free:

curl https://syntology.ai/api/ran/2306.05422.json

A badge for a README (the split and the date, never a ratio):

[![Syntology run record](https://syntology.ai/api/ran/2306.05422.svg)](https://syntology.ai/paper/2306.05422)

Connect an agent — have() is free.