SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2106.00677 · ICCV · 2021

Bootstrap Your Own Correspondences

Justin Johnson, Mohamed El Banani

arXiv · PDF · Open in the Atlas

Code that ran

3 of 9 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 3 have no contract verdict; 0 failures of our instrument; 6 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/2106.00677.json · badge: /api/ran/2106.00677.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
mbanani/byoc canonical (classifier label)
classifier basis: census
3 ran · 6 no recorded run
FunctionStatusWhere it lives
calculate_ratio_test Ran
ran on a synthesized input
mbanani/byoc/byoc/models/correspondence.py
code served (permissive licence) · get_code("1c65d2c53b3b926b")
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 3, 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

torch.manual_seed(0)

# Create a small tensor of shape (N, P, 2) representing cosine distances
# between points and their 2 nearest neighbors
# N=2 (batch size), P=4 (number of points), 2 (distances to 2 nearest neighbors)
dists = torch.rand(2, 4, 2) * 0.5  # Keep values small (between 0 and 0.5) to be valid distances

ARGS = {
    "dists": dists
}
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 3, dtype family float — no output shape is stated: observe_signature.py could not observe it (import:OSError)
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
evaluate_correspondances Ran
ran on a synthesized input
mbanani/byoc/byoc/utils/metrics.py
code served (permissive licence) · get_code("7125de3e5cc61b12")
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

torch.manual_seed(0)

# Define K as a small stand-in class that implements the matrix operations used
class CameraMatrix:
    def __init__(self, K_tensor):
        self.K = K_tensor
    
    def inverse(self):
        """Return an object that has a bmm method"""
        K_inv = torch.inverse(self.K)
        return BatchMatrixWrapper(K_inv)
    
    def bmm(self, x):
        """Batch matrix multiply"""
        return self.K.bmm(x)

class BatchMatrixWrapper:
    def __init__(self, matrix):
        self.matrix = matrix
    
    def bmm(self, x):
        """Batch matrix multiply"""
        return self.matrix.bmm(x)

# Set up tensor dimensions
batch_size = 2
num_matches = 4

# pixel_loc_0: batch_size x 2 x num_matches (LongTensor)
pixel_loc_0 = torch.randint(0, 10, (batch_size, 2, num_matches), dtype=torch.long)

# pixel_loc_1: batch_size x 2 x num_matches (LongTensor)
pixel_loc_1 = torch.randint(0, 10, (batch_size, 2, num_matches), dtype=torch.long)

# depth_0: batch_size x 1 x num_matches (FloatTensor), must be positive
depth_0 = torch.rand(batch_size, 1, num_matches) + 0.5

# K: camera intrinsic matrix, needs to support inverse() and bmm()
# Standard 3x3 camera intrinsic matrix
K_matrix = torch.tensor([
    [[500.0, 0.0, 320.0],
     [0.0, 500.0, 240.0],
     [0.0, 0.0, 1.0]],
    [[500.0, 0.0, 320.0],
     [0.0, 500.0, 240.0],
     [0.0, 0.0, 1.0]]
])
K = CameraMatrix(K_matrix)

# Rt: batch_size x 3 x 4, rotation and translation matrix
Rt = torch.zeros(batch_size, 3, 4)
# Set rotation part to identity-like
Rt[:, 0, 0] = 1.0
Rt[:, 1, 1] = 1.0
Rt[:, 2, 2] = 1.0
# Set translation part (column 3)
Rt[:, :, 3] = torch.randn(batch_size, 3)

ARGS = {
    'pixel_loc_0': pixel_loc_0,
    'pixel_loc_1': pixel_loc_1,
    'depth_0': depth_0,
    'K': K,
    'Rt': Rt
}
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
get_topk_matches Ran
ran on a synthesized input
mbanani/byoc/byoc/models/correspondence.py
code served (permissive licence) · get_code("613a3a3a21f9a4dd")
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 3, 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

torch.manual_seed(0)

# Construct arguments for get_topk_matches
# The function expects:
# - dists: tensor of distances, shape (batch, num_candidates)
# - idx: tensor of indices, shape (batch, num_candidates)
# - num_corres: int, number of correspondences to retrieve

batch_size = 2
num_candidates = 4

# Create dists tensor: shape (batch_size, num_candidates)
# Contains distance values
dists = torch.randn(batch_size, num_candidates)

# Create idx tensor: shape (batch_size, num_candidates)
# Contains indices that will be gathered
idx = torch.arange(num_candidates).unsqueeze(0).expand(batch_size, -1)

# Number of correspondences to retrieve
num_corres = 3

ARGS = {
    "dists": dists,
    "idx": idx,
    "num_corres": num_corres,
}
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 3, 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
align Not yet run
unverified (harvested, no recorded run)
mbanani/byoc/byoc/models/alignment.py
code served (permissive licence) · get_code("2508b020295d7823")
How it ran

no invocation recorded.

build_dataset Not yet run
unverified (harvested, no recorded run)
mbanani/byoc/byoc/datasets/builder.py
code served (permissive licence) · get_code("63c1df24808bc1ae")
How it ran

no invocation recorded.

build_model Not yet run
unverified (harvested, no recorded run)
mbanani/byoc/byoc/models/builder.py
code served (permissive licence) · get_code("fca9a8934386530f")
How it ran

no invocation recorded.

evaluate_3d_correspondances Not yet run
unverified (harvested, no recorded run)
mbanani/byoc/byoc/utils/metrics.py
code served (permissive licence) · get_code("c65a0c6c9123e7f5")
How it ran

no invocation recorded.

paired_svd Not yet run
unverified (harvested, no recorded run)
mbanani/byoc/byoc/models/alignment.py
code served (permissive licence) · get_code("56c4a83a8b6469be")
How it ran

no invocation recorded.

randomized_weighted_procrustes Not yet run
unverified (harvested, no recorded run)
mbanani/byoc/byoc/models/alignment.py
code served (permissive licence) · get_code("5ddb3c71c6feae1f")
How it ran

no invocation recorded.

Also present in other papers' code

Of 9 samples lifted from this paper's linked repositories, 9 carry a function identity and 0 do not. Of the 9 distinct functions with an identity, 2 are also present in other papers whose code we hold (2 with at least one paper dated earlier, 0 the same month, 0 later) and 7 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
paired_svd 1 other paper
1 dated earlier
2102.11870 (2021-02)
UnsupervisedR&R: Unsupervised Point Cloud Registration via Differentiable Rendering
randomized_weighted_procrustes 1 other paper
1 dated earlier
2102.11870 (2021-02)
UnsupervisedR&R: Unsupervised Point Cloud Registration via Differentiable Rendering

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

Geometric feature extraction is a crucial component of point cloud registration pipelines. Recent work has demonstrated how supervised learning can be leveraged to learn better and more compact 3D features. However, those approaches' reliance on ground-truth annotation limits their scalability. We propose BYOC: a self-supervised approach that learns visual and geometric features from RGB-D video without relying on ground-truth pose or correspondence. Our key observation is that randomly-initialized CNNs readily provide us with good correspondences; allowing us to bootstrap the learning of both visual and geometric features. Our approach combines classic ideas from point cloud registration with more recent representation learning approaches. We evaluate our approach on indoor scene datasets and find that our method outperforms traditional and learned descriptors, while being competitive with current state-of-the-art supervised approaches.

For agents

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

get_harvested_code_for_paper("2106.00677")
get_code_for_paper("2106.00677")
have("2106.00677")

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

curl https://syntology.ai/api/ran/2106.00677.json

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

[![Syntology run record](https://syntology.ai/api/ran/2106.00677.svg)](https://syntology.ai/paper/2106.00677)

Connect an agent — have() is free.