SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 1906.11883 · 2019

Unsupervised Learning of Object Keypoints for Perception and Control

arXiv · PDF · Open in the Atlas

Code that ran

10 of 11 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 6 have no contract verdict; 4 failures of our instrument; 1 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/1906.11883.json · badge: /api/ran/1906.11883.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
DuaneNielsen/keypoints community (classifier label)
classifier basis not recorded on the link
6 ran · 1 no recorded run
DerrickGXD/unsupervised-keypoints reimplementation (classifier label)
classifier basis: classifier_2025
4 ran · 0 no recorded run
FunctionStatusWhere it lives
channels_to_frame Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
DerrickGXD/unsupervised-keypoints/TrainDataset.py
pointer only (licence: NONE) · get_code("b23a47ea77508e05")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, DerrickGXD/unsupervised-keypoints 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.

get_gaussian_mean Ran
ran; our contract draft was wrong, not the code
DerrickGXD/unsupervised-keypoints/Models.py
pointer only (licence: NONE) · get_code("6cd6f0b3a526afba")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, DerrickGXD/unsupervised-keypoints 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.

isPD Ran
ran on a synthesized input
DuaneNielsen/keypoints/keypoints/higgham.py
code served (permissive licence) · get_code("c4eadb0bceb3178a")
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 bool, ndim 0, dtype family float, 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

torch.manual_seed(0)

# isPD expects a tensor A and attempts to compute its Cholesky decomposition
# A must be a square, symmetric positive-definite matrix for the function to return True
# Let's create a small positive-definite matrix

# Create a small positive-definite matrix by constructing A = X^T @ X
X = torch.randn(4, 4)
A = X.T @ X  # This is guaranteed to be positive-definite (or semi-definite)

# Ensure it's strictly positive-definite by adding to diagonal
A = A + torch.eye(4) * 0.1

ARGS = {
    "A": A
}
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 bool, ndim 0, dtype family float, 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
loss_alex Ran
ran; our contract draft was wrong, not the code
DerrickGXD/unsupervised-keypoints/Models.py
pointer only (licence: NONE) · get_code("cd62068519eb2f35")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, DerrickGXD/unsupervised-keypoints 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.

marginal_logsoftmax Ran
ran on a synthesized input
DuaneNielsen/keypoints/keypoints/models/functional.py
code served (permissive licence) · get_code("c7c19dfcdf23b96f")
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, shape (2, 4, 5) 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.functional as F

torch.manual_seed(0)

# The function expects:
# - heatmap: a tensor that will be reduced via torch.mean along a dimension,
#            then log_softmax applied on dim=2
# - dim: an integer specifying which dimension to reduce via mean

# Looking at the function body:
# - heatmap needs at least 3 dimensions (since log_softmax is applied on dim=2)
# - dim should be an integer specifying the reduction dimension
# - After mean reduction along dim, the result should have at least 3 dimensions
#   for log_softmax(dim=2) to work

# Create a heatmap tensor: shape (2, 3, 4, 5)
# This has 4 dimensions, so we can reduce along different dims
# If we reduce along dim=1, we get shape (2, 4, 5), then log_softmax on dim=2 works
heatmap = torch.randn(2, 3, 4, 5)
dim = 1

ARGS = {
    "heatmap": heatmap,
    "dim": dim
}
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, shape (2, 4, 5) — shape (2, 4, 5) 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
marginal_softmax Ran
ran on a synthesized input
DuaneNielsen/keypoints/keypoints/models/functional.py
code served (permissive licence) · get_code("dc04488a5d9bd28b")
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, shape (2, 3, 5) 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.functional as F

torch.manual_seed(0)

# Create a small heatmap tensor
# Shape: (batch=2, height=4, width=3, channels=5)
# This allows us to test different dim values
heatmap = torch.randn(2, 4, 3, 5)

# dim parameter - which dimension to take mean over
# The function takes mean over this dim, then applies softmax over dim=2
# So we need to choose a dim that exists in the heatmap
dim = 1

ARGS = {
    "heatmap": heatmap,
    "dim": dim
}
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, shape (2, 3, 5) — shape (2, 3, 5) 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
np_isPD Ran
ran on a synthesized input
DuaneNielsen/keypoints/keypoints/higgham.py
code served (permissive licence) · get_code("92afecac079e4329")
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 bool, ndim 0, dtype family float, 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 numpy as np

# The function np_isPD takes a matrix B and checks if it's positive-definite
# using Cholesky decomposition. The function expects a numpy array.

# Create a positive-definite matrix for testing
# A simple way is to create a matrix A and compute A @ A.T
np.random.seed(0)
A = np.random.randn(4, 4)
B = A @ A.T  # This guarantees positive-definiteness

ARGS = {
    'B': B
}
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 bool, ndim 0, dtype family float, 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
np_nearestPD Ran
ran on a synthesized input
DuaneNielsen/keypoints/keypoints/higgham.py
code served (permissive licence) · get_code("4902b87f9d94f562")
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 (4, 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 numpy as np
from numpy import linalg as la

torch.manual_seed(0)

# Create a small non-positive-definite matrix to test np_nearestPD
# Start with a random matrix and make it symmetric but not PD
A = np.random.randn(4, 4)
A = (A + A.T) / 2  # Make it symmetric
# Perturb it to ensure it's not positive-definite
A = A - np.eye(4) * 5

ARGS = {
    "A": A
}
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 2, dtype family float, shape (4, 4) — shape (4, 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
prob_to_keypoints Ran
ran on a synthesized input
DuaneNielsen/keypoints/keypoints/models/functional.py
code served (permissive licence) · get_code("b80d8ae1fc2741ec")
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, 1) 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)

# prob should be a tensor with shape (batch, channels, length)
# where the function will compute a weighted sum along dimension 2
# Using batch=2, channels=1, length=4
prob = torch.rand(2, 1, 4)

# length parameter - the number of positions along the ruler
length = 4

ARGS = {
    "prob": prob,
    "length": length,
}
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, 1) — shape (2, 1) 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
rgba_to_rgb Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
DerrickGXD/unsupervised-keypoints/Models.py
pointer only (licence: NONE) · get_code("6c2bf41ef195a80a")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, DerrickGXD/unsupervised-keypoints 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.

config Not yet run
unverified (harvested, no recorded run)
DuaneNielsen/keypoints/keypoints/config.py
code served (permissive licence) · get_code("296edbfe5d4adb67")
How it ran

no invocation recorded.

Also present in other papers' code

Of 11 samples lifted from this paper's linked repositories, 11 carry a function identity and 0 do not. Of the 11 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 11 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

The study of object representations in computer vision has primarily focused on developing representations that are useful for image classification, object detection, or semantic segmentation as downstream tasks. In this work we aim to learn object representations that are useful for control and reinforcement learning (RL). To this end, we introduce Transporter, a neural network architecture for discovering concise geometric object representations in terms of keypoints or image-space coordinates. Our method learns from raw video frames in a fully unsupervised manner, by transporting learnt image features between video frames using a keypoint bottleneck. The discovered keypoints track objects and object parts across long time-horizons more accurately than recent similar methods. Furthermore, consistent long-term tracking enables two notable results in control domains -- (1) using the keypoint co-ordinates and corresponding image features as inputs enables highly sample-efficient reinforcement learning; (2) learning to explore by controlling keypoint locations drastically reduces the search space, enabling deep exploration (leading to states unreachable through random action exploration) without any extrinsic rewards.

For agents

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

get_harvested_code_for_paper("1906.11883")
get_code_for_paper("1906.11883")
have("1906.11883")

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

curl https://syntology.ai/api/ran/1906.11883.json

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

[![Syntology run record](https://syntology.ai/api/ran/1906.11883.svg)](https://syntology.ai/paper/1906.11883)

Connect an agent — have() is free.