SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2204.04153 · ECCV · 2022

Particle Video Revisited: Tracking Through Occlusions Using Point Trajectories

Katerina Fragkiadaki, Zhaoyuan Fang, Adam Harley

arXiv · PDF · Open in the Atlas

Code that ran

5 of 5 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (0 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 5 have no contract verdict; 0 failures of our instrument; 0 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/2204.04153.json · badge: /api/ran/2204.04153.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
aharley/pips canonical (classifier label)
classifier basis not recorded on the link
5 ran · 0 no recorded run
FunctionStatusWhere it lives
fetch_optimizer Ran
ran on a synthesized input
aharley/pips/train2.py
code served (permissive licence) · get_code("db83affa17361b10")
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 tuple, ndim 1, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim 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)

# Create a small model with parameters to optimize
class SmallModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(4, 3)
    
    def forward(self, x):
        return self.linear(x)

model = SmallModel()

ARGS = {
    'lr': 0.001,
    'wdecay': 0.01,
    'epsilon': 1e-8,
    'num_steps': 1000,
    'params': model.parameters()
}
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 tuple, ndim 1, dtype family float — a tuple output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
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
flood_fill_hull Ran
ran on a synthesized input
aharley/pips/make_occlusions.py
code served (permissive licence) · get_code("768d8abfae882911")
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 tuple. 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 scipy.spatial

# Create a simple binary image with some non-zero points
# The function expects an image where True/non-zero pixels define the region
# We need at least 4 points in 2D (or d+1 points in d dimensions) for ConvexHull
image = np.zeros((5, 5), dtype=bool)
# Set some points to True to form a region
image[1, 1] = True
image[1, 3] = True
image[3, 1] = True
image[3, 3] = True
image[2, 2] = True

ARGS = {
    'image': image
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, scipy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple — a tuple output has no array shape of its own; none is 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
load Ran
ran on a synthesized input
aharley/pips/saverloader.py
code served (permissive licence) · get_code("a9f4f1b5557431c4")
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 int, ndim 0, dtype family int, shape () as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape 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
import os
import tempfile
import shutil

torch.manual_seed(0)

# Create a minimal model
class MinimalModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(4, 3)
    
    def forward(self, x):
        return self.linear(x)

# Create a minimal optimizer
model = MinimalModel()
optimizer = torch.optim.SGD(model.parameters(), lr=0.01)

# Create a minimal scheduler
scheduler = torch.optim.lr_scheduler.StepLR(optimizer, step_size=1)

# Create a minimal EMA model
model_ema = MinimalModel()

# Create a temporary checkpoint directory with a valid checkpoint file
ckpt_dir = tempfile.mkdtemp()

# Create a sample checkpoint
checkpoint = {
    'model_state_dict': model.state_dict(),
    'optimizer_state_dict': optimizer.state_dict(),
    'scheduler_state_dict': scheduler.state_dict(),
    'ema_model_state_dict': model_ema.state_dict(),
}

ckpt_path = os.path.join(ckpt_dir, 'model-000000001.pth')
torch.save(checkpoint, ckpt_path)

# Set up arguments
ARGS = {
    'ckpt_dir': ckpt_dir,
    'model': model,
    'optimizer': optimizer,
    'scheduler': scheduler,
    'model_ema': model_ema,
    'step': 0,
    'model_name': 'model',
    'ignore_load': None,
}
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 int, ndim 0, dtype family int, shape () — shape () as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
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
readImage Ran
ran on a synthesized input
aharley/pips/flyingthingsdataset.py
code served (permissive licence) · get_code("8dc07cad5c147d48")
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 ndarray, ndim 3, dtype family float, shape (4, 3, 3) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape 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 tempfile
import os

# Create a temporary PFM file for testing
def create_test_pfm(filename, height=4, width=3, color=True):
    """Create a minimal valid PFM file."""
    with open(filename, 'wb') as f:
        # Write header
        if color:
            f.write(b'PF\n')
        else:
            f.write(b'Pf\n')
        
        # Write dimensions
        f.write(f'{width} {height} \n'.encode('ascii'))
        
        # Write scale (positive = big-endian)
        f.write(b'1.0\n')
        
        # Write data
        if color:
            data = np.random.rand(height, width, 3).astype(np.float32)
        else:
            data = np.random.rand(height, width).astype(np.float32)
        
        # Flip vertically before writing (PFM convention)
        data = np.flipud(data)
        data.tofile(f)

# Create a temporary directory and PFM file
temp_dir = tempfile.mkdtemp()
pfm_file = os.path.join(temp_dir, 'test.pfm')
create_test_pfm(pfm_file, height=4, width=3, color=True)

ARGS = {
    'name': pfm_file
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
imageio, numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 3, dtype family float, shape (4, 3, 3) — shape (4, 3, 3) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
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
readPFM Ran
ran on a synthesized input
aharley/pips/flyingthingsdataset.py
code served (permissive licence) · get_code("d4c1dfe6e2a09277")
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 ndarray, ndim 3, dtype family float, shape (3, 4, 3) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape 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 tempfile
import os

# Create a temporary PFM file for testing
# PFM format: 
# - Header: "PF" (color) or "Pf" (grayscale)
# - Dimensions: "width height\n"
# - Scale: "scale\n" (negative means little-endian)
# - Data: binary floats in row-major order (flipped vertically)

def create_test_pfm():
    """Create a small test PFM file"""
    width, height = 4, 3
    
    # Create simple test data (height x width x 3 for color)
    data = np.random.randn(height, width, 3).astype(np.float32)
    
    # PFM files store data flipped vertically
    data_flipped = np.flipud(data)
    
    # Create temporary file
    fd, filepath = tempfile.mkstemp(suffix='.pfm')
    os.close(fd)
    
    with open(filepath, 'wb') as f:
        # Write header
        f.write(b'PF\n')
        # Write dimensions
        f.write(f'{width} {height}\n'.encode('ascii'))
        # Write scale (negative means little-endian)
        f.write(b'-1.0\n')
        # Write data in little-endian format
        f.write(data_flipped.astype('<f').tobytes())
    
    return filepath

pfm_file_path = create_test_pfm()

ARGS = {
    'file': pfm_file_path
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 3, dtype family float, shape (3, 4, 3) — shape (3, 4, 3) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
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

Also present in other papers' code

Of 5 samples lifted from this paper's linked repositories, 5 carry a function identity and 0 do not. Of the 5 distinct functions with an identity, 0 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 0 later) and 5 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.

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

Tracking pixels in videos is typically studied as an optical flow estimation problem, where every pixel is described with a displacement vector that locates it in the next frame. Even though wider temporal context is freely available, prior efforts to take this into account have yielded only small gains over 2-frame methods. In this paper, we revisit Sand and Teller's "particle video" approach, and study pixel tracking as a long-range motion estimation problem, where every pixel is described with a trajectory that locates it in multiple future frames. We re-build this classic approach using components that drive the current state-ofthe-art in flow and object tracking, such as dense cost maps, iterative optimization, and learned appearance updates. We train our models using long-range amodal point trajectories mined from existing optical flow data that we synthetically augment with multi-frame occlusions. We test our approach in trajectory estimation benchmarks and in keypoint label propagation tasks, and compare favorably against state-of-the-art optical flow and feature tracking methods.

For agents

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

get_harvested_code_for_paper("2204.04153")
get_code_for_paper("2204.04153")
have("2204.04153")

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

curl https://syntology.ai/api/ran/2204.04153.json

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

[![Syntology run record](https://syntology.ai/api/ran/2204.04153.svg)](https://syntology.ai/paper/2204.04153)

Connect an agent — have() is free.