SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2303.17778 · ICCV · 2023

CROSSLOC3D: Aerial-Ground Cross-Source 3D Place Recognition

Dinesh Manocha, Xijun Wang, Tianrui Guan, Jing Liang, Adarsh Sathyamoorthy, Aswath Muthuselvam, Montana Hoover, Damon Conover

arXiv · PDF · Open in the Atlas

Code that ran

11 of 15 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (9 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 10 have no contract verdict; 1 failure of our instrument; 4 no recorded run), measured 2026-11-22; 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/2303.17778.json · badge: /api/ran/2303.17778.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
rayguan97/crossloc3d marked official in Papers with Code
Papers with Code archive
11 ran · 4 no recorded run
FunctionStatusWhere it lives
DiffTransformerBlock Ran
ran on a synthesized input · constructed an object; no result was computed
rayguan97/crossloc3d/models/ours/model.py
code served (permissive licence) · get_code("49a28a91da28cd6a")
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, DiffTransformerBlock: 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
import math

torch.manual_seed(0)

# Define a minimal config class
class Config:
    def __init__(self):
        self.global_channels = 8
        self.time_dim = 4
        self.learned_sinusoidal_cond = False
        self.num_attn_layers = 4
        self.local_channels = 2
        self.num_heads = 2
        self.num_centers = [2, 2, 2, 2]

# Create config instance
cfg = Config()

# q_size and step_size should multiply to num_attn_layers
q_size = [2, 2]  # len(q_size) = 2
step_size = 1    # len(q_size) * step_size = 2 * 1 = 2 (but cfg.num_attn_layers = 4)

# Actually, let's fix this so they match
q_size = [2]     # len(q_size) = 1
step_size = 4    # len(q_size) * step_size = 1 * 4 = 4, matches cfg.num_attn_layers

ARGS = {
    'cfg': cfg,
    'q_size': q_size,
    'step_size': step_size
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
einops, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type DiffTransformerBlock — 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 DiffTransformerBlock 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
GatingContext Ran
ran on a synthesized input · constructed an object; no result was computed
rayguan97/crossloc3d/models/ours/model.py
code served (permissive licence) · get_code("25dfda388912d104")
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, GatingContext: 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
import math

torch.manual_seed(0)

ARGS = {
    "dim": 4,
    "add_batch_norm": True,
}
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 GatingContext — 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 GatingContext 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
GeM Ran
ran on a synthesized input · constructed an object; no result was computed
rayguan97/crossloc3d/models/ours/model.py
code served (permissive licence) · get_code("039a0681e332e8c7")
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, GeM: 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)

class Config:
    def __init__(self, p=3.0, eps=1e-6):
        self.p = p
        self.eps = eps

cfg = Config(p=3.0, eps=1e-6)

ARGS = {
    "cfg": cfg
}
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 GeM — 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 GeM 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
MAC Ran
ran on a synthesized input · constructed an object; no result was computed
rayguan97/crossloc3d/models/ours/model.py
code served (permissive licence) · get_code("a137e6ae7a2306dc")
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, MAC: 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)

class Config:
    def __init__(self):
        self.hidden_dim = 64
        self.num_layers = 2

cfg = Config()

ARGS = {
    'cfg': cfg
}
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 MAC — 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 MAC 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
NetVLADLoupe Ran
ran on a synthesized input · constructed an object; no result was computed
rayguan97/crossloc3d/models/ours/model.py
code served (permissive licence) · get_code("d9300c949a291ace")
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, NetVLADLoupe: 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
import math

torch.manual_seed(0)

# Set up parameters for NetVLADLoupe
feature_size = 4
cluster_size = 3
output_dim = 5
gating = True
add_batch_norm = True

ARGS = {
    "feature_size": feature_size,
    "cluster_size": cluster_size,
    "output_dim": output_dim,
    "gating": gating,
    "add_batch_norm": add_batch_norm,
}
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 NetVLADLoupe — 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 NetVLADLoupe 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
NetVladWrapper Ran
ran on a synthesized input · constructed an object; no result was computed
rayguan97/crossloc3d/models/ours/model.py
code served (permissive licence) · get_code("44a4c9c114054f04")
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, NetVladWrapper: 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
import math

torch.manual_seed(0)

class Config:
    def __init__(self):
        self.in_channels = 4
        self.out_channels = 8
        self.cluster_size = 3
        self.gating = True
        self.add_bn = True

ARGS = {
    'cfg': Config()
}
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 NetVladWrapper — 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 NetVladWrapper 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
RandomOrLearnedSinusoidalPosEmb Ran
ran on a synthesized input · constructed an object; no result was computed
rayguan97/crossloc3d/models/ours/model.py
code served (permissive licence) · get_code("4611d53e8f11052e")
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, RandomOrLearnedSinusoidalPosEmb: 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
import math
from einops import rearrange, reduce, repeat

torch.manual_seed(0)

# The function being tested is RandomOrLearnedSinusoidalPosEmb
# It takes two parameters in __init__: dim and is_random

ARGS = {
    'dim': 4,
    'is_random': False
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
einops, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type RandomOrLearnedSinusoidalPosEmb — 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 RandomOrLearnedSinusoidalPosEmb 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
SPoC Ran
ran on a synthesized input · constructed an object; no result was computed
rayguan97/crossloc3d/models/ours/model.py
code served (permissive licence) · get_code("004ffb97f34d150e")
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, SPoC: 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)

# Create a minimal config dict
cfg = {}

# Create input tensor: (batch=2, feature_dim=4, sequence_length=3)
# The forward method calls torch.mean(x, dim=-1, keepdim=False)
# so we need a 3D tensor where dim=-1 will be reduced
x = torch.randn(2, 4, 3)

ARGS = {
    'cfg': cfg,
}
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 SPoC — 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 SPoC 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
SinusoidalPosEmb Ran
ran on a synthesized input
rayguan97/crossloc3d/models/ours/model.py
code served (permissive licence) · get_code("f97842720e968c4b")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, class construct-then-call prompt, model not recorded) and returned Tensor, ndim 2, dtype family float, shape (2, 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 torch
import torch.nn as nn
import math

class SinusoidalPosEmb(nn.Module):

    def __init__(self, dim):
        super().__init__()
        assert dim % 2 == 0
        self.dim = dim

    def forward(self, x):
        device = x.device
        half_dim = self.dim // 2
        emb = math.log(10000) / (half_dim - 1)
        emb = torch.exp(torch.arange(half_dim, device=device) * -emb)
        emb = x[:, None] * emb[None, :]
        emb = torch.cat((emb.sin(), emb.cos()), dim=-1)
        return emb


INIT_ARGS = {
    'dim': 4
}

ARGS = {
    'x': torch.randn(2)
}

if __name__ == '__main__':
    model = SinusoidalPosEmb(**INIT_ARGS)
    output = model(**ARGS)
    print(f"Input shape: {ARGS['x'].shape}")
    print(f"Output shape: {output.shape}")
    print(f"Output:\n{output}")
Setup block drafted by
a language model, class construct-then-call prompt, model not recorded
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 2, dtype family float, shape (2, 4) — shape (2, 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: torch and numpy seeded to 0 by the harness immediately before the class was constructed from INIT_ARGS (fixture_check.py)
TCenterAttentionLayer Ran
ran on a synthesized input · constructed an object; no result was computed
rayguan97/crossloc3d/models/ours/model.py
code served (permissive licence) · get_code("7b1d62857496e935")
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, TCenterAttentionLayer: 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
import math

torch.manual_seed(0)

# Define ARGS dict with all parameters for TCenterAttentionLayer.__init__
ARGS = {
    'global_channels': 8,
    'num_centers': 4,
    'local_channels': 0,
    'num_heads': 2
}

# Verify the setup works by instantiating the module
model = nn.Module()
layer = nn.Module.__new__(type('TCenterAttentionLayer', (nn.Module,), {}))

# Now test forward pass setup
bs = 2
global_channels = ARGS['global_channels']
npts = 3

local_x = torch.randn(bs, global_channels, npts)
t_embed = torch.randn(bs, global_channels)
f_p = torch.randn(bs, global_channels)

# These would be used in forward call
FORWARD_ARGS = {
    'local_x': local_x,
    't_embed': t_embed,
    'f_p': f_p
}

ARGS
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 TCenterAttentionLayer — 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 TCenterAttentionLayer 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
l2norm Ran
ran; our contract draft was wrong, not the code
rayguan97/crossloc3d/models/ours/model.py
code served (permissive licence) · get_code("8deabf7913c20741")
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 Tensor, ndim 2, dtype family float, shape (2, 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 torch
import torch.nn as nn

torch.manual_seed(0)

# Create a small tensor with shape (2, 4)
# dim=0 means we'll normalize along the first dimension
inp = torch.randn(2, 4)

ARGS = {
    "inp": inp,
    "dim": 0
}
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 Tensor, ndim 2, dtype family float, shape (2, 4) — shape (2, 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
ECALayer Not yet run
unverified (harvested, no recorded run)
rayguan97/crossloc3d/models/ours/model.py
code served (permissive licence) · get_code("5a185e87c545019a")
How it ran

no invocation recorded.

GatedInceptionBlock Not yet run
unverified (harvested, no recorded run)
rayguan97/crossloc3d/models/ours/model.py
code served (permissive licence) · get_code("de332b4043605765")
How it ran

no invocation recorded.

Ours Not yet run
unverified (harvested, no recorded run)
rayguan97/crossloc3d/models/ours/model.py
code served (permissive licence) · get_code("05dd8633cd267827")
How it ran

no invocation recorded.

Ours3DFPN Not yet run
unverified (harvested, no recorded run)
rayguan97/crossloc3d/models/ours/model.py
code served (permissive licence) · get_code("2a6a78f87852fb30")
How it ran

no invocation recorded.

Also present in other papers' code

Of 15 samples lifted from this paper's linked repositories, 15 carry a function identity and 0 do not. Of the 15 distinct functions with an identity, 1 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 1 later) and 14 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
RandomOrLearnedSinusoidalPosEmb
brick of this paper's assembly: helper
1 other paper
1 later
2403.00939 (2024-03)
G3DR: Generative 3D Reconstruction in ImageNet

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

We present CROSSLOC3D, a novel 3D place recognition method that solves a large-scale point matching problem in a cross-source setting. Cross-source point cloud data corresponds to point sets captured by depth sensors with different accuracies or from different distances and perspectives. We address the challenges in terms of developing 3D place recognition methods that account for the representation gap between points captured by different sources. Our method handles cross-source data by utilizing multigrained features and selecting convolution kernel sizes that correspond to most prominent features. Inspired by the diffusion models, our method uses a novel iterative refinement process that gradually shifts the embedding spaces from different sources to a single canonical space for better metric learning. In addition, we present CS-CAMPUS3D, the first 3D aerial-ground cross-source dataset consisting of point cloud data from both aerial and ground LiDAR scans. The point clouds in CS-CAMPUS3D have representation gaps and other features like different views, point densities, and noise patterns. We show that our CROSSLOC3D algorithm can achieve an improvement of 4.74% -15.37% in terms of the top 1 average recall on our CS-CAMPUS3D benchmark and achieves performance comparable to state-of-theart 3D place recognition method on the Oxford RobotCar. The code and CS-CAMPUS3D benchmark will be available at github.com/rayguan97/crossloc3d .

For agents

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

get_harvested_code_for_paper("2303.17778")
get_code_for_paper("2303.17778")
have("2303.17778")

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

curl https://syntology.ai/api/ran/2303.17778.json

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

[![Syntology run record](https://syntology.ai/api/ran/2303.17778.svg)](https://syntology.ai/paper/2303.17778)

Connect an agent — have() is free.