SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2007.09919 · 2020

Robust Tracking against Adversarial Attacks

arXiv · PDF · Open in the Atlas

Code that ran

6 of 6 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; 2 honoured the contract we drafted, 0 violated it and 0 have no contract verdict; 4 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/2007.09919.json · badge: /api/ran/2007.09919.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
joshuajss/RTAA canonical (classifier label)
classifier basis: classifier_2025
3 ran · 0 no recorded run
vision-sjtu/rtaa community (classifier label)
classifier basis not recorded on the link
3 ran · 0 no recorded run
FunctionStatusWhere it lives
generate_anchor Ran
ran; our contract draft was wrong, not the code
joshuajss/RTAA/DaSiamRPN/code/run_SiamRPN.py
code served (permissive licence) · get_code("ffd959a18529f028")
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 2, dtype family float, shape (225, 4) 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

ARGS = {
    'total_stride': 16,
    'scales': [0.5, 1.0, 2.0],
    'ratios': [0.5, 1.0, 2.0],
    'score_size': 5
}
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 2, dtype family float, shape (225, 4) — shape (225, 4) 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
im_to_numpy Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
vision-sjtu/rtaa/DaSiamRPN/code/utils.py
code served (permissive licence) · get_code("fec5535bc7550e01")
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, 5, 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 torch
import numpy as np

torch.manual_seed(0)

# Create a small test image tensor with shape (C, H, W) = (3, 4, 5)
# This will be transposed to (H, W, C) = (4, 5, 3)
img_tensor = torch.randn(3, 4, 5)

ARGS = {
    'img': img_tensor
}
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 ndarray, ndim 3, dtype family float, shape (4, 5, 3) — shape (4, 5, 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
to_numpy Ran
ran, honoured the contract we drafted
vision-sjtu/rtaa/DaSiamRPN/code/utils.py
pointer only (licence: MIT); the stored copy get_code serves, from SY-Xuan/IICS, is pointer only (licence: NONE) · get_code("b476623700945834")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (MIT), but get_code serves the stored copy, first harvested from SY-Xuan/IICS, and GitHub detected no licence in that repository when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

to_torch Ran
ran, honoured the contract we drafted
vision-sjtu/rtaa/DaSiamRPN/code/utils.py
pointer only (licence: MIT); the stored copy get_code serves, from liu-xb/GCMT, is pointer only (licence: NONE) · get_code("7aac4c51e0ceee97")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (MIT), but get_code serves the stored copy, first harvested from liu-xb/GCMT, and GitHub detected no licence in that repository when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

tracker_eval Ran
ran; our contract draft was wrong, not the code
joshuajss/RTAA/DaSiamRPN/code/run_SiamRPN.py
code served (permissive licence) · get_code("ac7f8214437637db")
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 torch
import torch.nn as nn
import numpy as np

torch.manual_seed(0)

class SimpleTracker(nn.Module):
    def __init__(self):
        super().__init__()
        self.conv = nn.Conv2d(3, 4, 1)
        self.score_conv = nn.Conv2d(3, 2, 1)
    
    def forward(self, x_crop):
        delta = self.conv(x_crop)
        score = self.score_conv(x_crop)
        return delta, score

class Params:
    def __init__(self):
        self.anchor = np.random.randn(400, 4)
        self.penalty_k = 0.04
        self.window_influence = 0.4
        self.lr = 0.3

net = SimpleTracker()
x_crop = torch.randn(1, 3, 20, 20)
target_pos = np.array([100.0, 100.0])
target_sz = np.array([50.0, 50.0])
window = np.random.randn(400)
scale_z = 1.0
p = Params()

ARGS = {
    'net': net,
    'x_crop': x_crop,
    'target_pos': target_pos,
    'target_sz': target_sz,
    'window': window,
    'scale_z': scale_z,
    'p': p
}
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 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
tracker_eval Ran
ran; our contract draft was wrong, not the code
joshuajss/RTAA/DaSiamRPN/code/run_attack.py
code served (permissive licence) · get_code("fd9e88b104e6b6f7")
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 torch
import torch.nn as nn
import numpy as np

torch.manual_seed(0)

# Define a minimal network that outputs delta and score tensors
class SimpleTracker(nn.Module):
    def __init__(self):
        super().__init__()
        self.conv1 = nn.Conv2d(3, 16, 1)
    
    def forward(self, x_crop):
        # x_crop shape: (1, 3, 31, 31) for example
        # delta output: (1, 4, height, width)
        # score output: (1, 2, height, width)
        batch_size = x_crop.shape[0]
        height = width = 5  # Small spatial dimensions
        delta = torch.randn(batch_size, 4, height, width)
        score = torch.randn(batch_size, 2, height, width)
        return delta, score

net = SimpleTracker()

# x_crop: input tensor (batch, channels, height, width)
x_crop = torch.randn(1, 3, 31, 31)

# target_pos: current target position [x, y]
target_pos = np.array([100.0, 100.0])

# target_sz: current target size [width, height]
target_sz = np.array([50.0, 50.0])

# window: penalty window (1D array matching number of anchors)
num_anchors = 25  # 5x5 spatial grid
window = np.hanning(num_anchors)

# scale_z: scale factor
scale_z = 1.5

# p: parameter object with required attributes
class Params:
    def __init__(self):
        # anchor shape: (num_anchors, 4) for [x, y, w, h]
        self.anchor = np.random.randn(num_anchors, 4)
        self.penalty_k = 0.05
        self.window_influence = 0.4
        self.lr = 0.3

p = Params()

# f: unused in function body, but required parameter
f = None

# gt: unused in function body, but required parameter
gt = None

# state: dictionary with image dimensions
state = {
    'im_w': 640,
    'im_h': 480
}

ARGS = {
    'net': net,
    'x_crop': x_crop,
    'target_pos': target_pos,
    'target_sz': target_sz,
    'window': window,
    'scale_z': scale_z,
    'p': p,
    'f': f,
    'gt': gt,
    'state': state
}
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 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

Also present in other papers' code

Of 6 samples lifted from this paper's linked repositories, 6 carry a function identity and 0 do not. Of the 6 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, 3 later) and 3 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
to_torch 9 other papers
2 dated earlier, 7 later · 3 only as an assembly brick
1912.08265 (2019-12)
Learning from Synthetic Animals
2003.02501 (2020-03)
Detecting Attended Visual Targets in Video
2103.11658 (2021-03)
Intra-Inter Camera Similarity for Unsupervised Person Re-Identification
and 6 more not named here
to_numpy 5 other papers
1 dated earlier, 4 later
2003.02501 (2020-03)
Detecting Attended Visual Targets in Video
2103.11658 (2021-03)
Intra-Inter Camera Similarity for Unsupervised Person Re-Identification
136670327 (2022 (year only))
Spatial-Separated Curve Rendering Network for Efficient and High-Resolution Image Harmonization
and 2 more not named here
im_to_numpy 3 other papers
1 dated earlier, 2 later
1912.08265 (2019-12)
Learning from Synthetic Animals
136670327 (2022 (year only))
Spatial-Separated Curve Rendering Network for Efficient and High-Resolution Image Harmonization
2308.09951 (2023-08)
Semantics Meets Temporal Correspondence: Self-supervised Object-centric Learning in Videos

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

While deep convolutional neural networks (CNNs) are vulnerable to adversarial attacks, considerably few efforts have been paid to construct robust deep tracking algorithms against adversarial attacks. Current studies on adversarial attack and defense mainly reside in a single image. In this work, we first attempt to generate adversarial examples on top of video sequences to improve the tracking robustness against adversarial attacks. To this end, we take temporal motion into consideration when generating lightweight perturbations over the estimated tracking results frame-by-frame. On one hand, we add the temporal perturbations into the original video sequences as adversarial examples to greatly degrade the tracking performance. On the other hand, we sequentially estimate the perturbations from input sequences and learn to eliminate their effect for performance restoration. We apply the proposed adversarial attack and defense approaches to state-of-the-art deep tracking algorithms. Extensive evaluations on the benchmark datasets demonstrate that our defense method not only eliminates the large performance drops caused by adversarial attacks, but also achieves additional performance gains when deep trackers are not under adversarial attacks.

For agents

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

get_harvested_code_for_paper("2007.09919")
get_code_for_paper("2007.09919")
have("2007.09919")

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

curl https://syntology.ai/api/ran/2007.09919.json

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

[![Syntology run record](https://syntology.ai/api/ran/2007.09919.svg)](https://syntology.ai/paper/2007.09919)

Connect an agent — have() is free.