SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2003.09572 · 2020

Monocular Real-time Hand Shape and Motion Capture using Multi-modal Data

arXiv · PDF · Open in the Atlas

Code that ran

10 of 14 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, 1 violated it and 8 have no contract verdict; 1 failure of our instrument; 4 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/2003.09572.json · badge: /api/ran/2003.09572.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
CalciferZh/minimal-hand canonical (classifier label)
classifier basis: census
7 ran · 2 no recorded run
vinnik-dmitry07/minimal-hand community (classifier label)
classifier basis not recorded on the link
3 ran · 2 no recorded run
FunctionStatusWhere it lives
calculate_auc Ran
ran on a synthesized input
CalciferZh/minimal-hand/plot.py
code served (permissive licence) · get_code("293235daae8170bd")
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 float, 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
import numpy as np

torch.manual_seed(0)

# Analyze the function:
# - xs: needs to be indexable with [-1] and [0], supports iteration with len()
#       elements support subtraction (xs[i+1] - xs[i])
# - ys: needs to be indexable, supports len() and iteration
#       elements support addition (ys[i] + ys[i+1]) and division

# Create simple list of x values (e.g., time points or positions)
xs = [0.0, 1.0, 2.0, 3.0]

# Create simple list of y values (e.g., function values at those points)
ys = [0.5, 1.0, 1.5, 2.0]

ARGS = {
    "xs": xs,
    "ys": ys,
}
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 float, 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
filter_nan_inf Ran
ran on a synthesized input
vinnik-dmitry07/minimal-hand/app1.py
code served (permissive licence) · get_code("3fcacade2a7a1d62")
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 1, dtype family float, shape (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 numpy as np

def filter_nan_inf(x):
    return x[np.isfinite(x)]

# Create a numpy array with some finite values, NaN, and inf
x = np.array([1.0, 2.0, np.nan, 4.0, np.inf, 6.0, -np.inf, 8.0])

ARGS = {
    "x": x
}
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 1, dtype family float, shape (5,) — shape (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
imresize Ran
ran on a synthesized input
CalciferZh/minimal-hand/utils.py
code served (permissive licence) · get_code("81cc19c4bc15b8bd")
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 int, shape (2, 2, 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 cv2
import numpy as np

# Create a small test image (height=3, width=4, channels=3)
# cv2 uses BGR format, so we need a 3D array
img = np.random.randint(0, 256, (3, 4, 3), dtype=np.uint8)

# Define target size as (width, height) tuple
size = (2, 2)

ARGS = {
    "img": img,
    "size": size
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
cv2, 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 int, shape (2, 2, 3) — shape (2, 2, 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
load_pkl Ran
ran; our contract draft was wrong, not the code
CalciferZh/minimal-hand/utils.py
code served (permissive licence) · get_code("7561717c5e734f6b")
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 dict. 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 pickle
import tempfile
import os

# Create a temporary pickle file with some test data
test_data = {
    'key1': 'value1',
    'key2': [1, 2, 3],
    'key3': {'nested': 'dict'}
}

# Write pickle file to temporary location
temp_dir = tempfile.gettempdir()
pickle_path = os.path.join(temp_dir, 'test_data.pkl')

with open(pickle_path, 'wb') as f:
    pickle.dump(test_data, f)

ARGS = {
    'path': pickle_path
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict 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
mano_to_mpii Ran
ran on a synthesized input
CalciferZh/minimal-hand/kinematics.py
code served (permissive licence) · get_code("98b35e2d80342219")
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 (21, 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

class MANOHandJoints:
    n_joints = 21
    labels = ['W', 'I0', 'I1', 'I2', 'M0', 'M1', 'M2', 'L0', 'L1', 'L2', 'R0', 'R1', 'R2', 'T0', 'T1', 'T2', 'I3', 'M3', 'L3', 'R3', 'T3']
    mesh_mapping = {16: 333, 17: 444, 18: 672, 19: 555, 20: 744}
    parents = [None, 0, 1, 2, 0, 4, 5, 0, 7, 8, 0, 10, 11, 0, 13, 14, 3, 6, 9, 12, 15]

class MPIIHandJoints:
    n_joints = 21
    labels = ['W', 'T0', 'T1', 'T2', 'T3', 'I0', 'I1', 'I2', 'I3', 'M0', 'M1', 'M2', 'M3', 'R0', 'R1', 'R2', 'R3', 'L0', 'L1', 'L2', 'L3']
    parents = [None, 0, 1, 2, 3, 0, 5, 6, 7, 0, 9, 10, 11, 0, 13, 14, 15, 0, 17, 18, 19]

# Create a sample MANO array with shape [21, 4, 3]
# 21 joints, 4 samples, 3 coordinates (x, y, z)
mano_data = np.random.randn(21, 4, 3).astype(np.float32)

ARGS = {
    "mano": mano_data
}
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 (21, 4, 3) — shape (21, 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
mpii_to_mano Ran
ran on a synthesized input
CalciferZh/minimal-hand/kinematics.py
code served (permissive licence) · get_code("91c388aacfa4de3c")
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 (21, 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

# The function expects mpii to be a numpy array with shape [21, ...] where 21 is the number of joints
# The function indexes along axis 0 and stacks results, so we need at least a 2D array
# Let's create a small example with shape [21, 3] representing 21 joints with 3D coordinates

mpii_data = np.random.randn(21, 3)

ARGS = {
    "mpii": mpii_data
}
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 (21, 3) — shape (21, 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_vec Ran
ran on a synthesized input
vinnik-dmitry07/minimal-hand/app1.py
code served (permissive licence) · get_code("2eeecc7deb1dcb9b")
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 dict. 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 numpy as np

torch.manual_seed(0)

# Analyze the function:
# def to_vec(pos):
#     return dict(zip(['x', 'y', 'z'], pos))
# 
# The function takes 'pos' as a parameter and zips it with keys ['x', 'y', 'z']
# 'pos' needs to be an iterable with at least 3 elements
# Could be a list, tuple, numpy array, torch tensor, etc.

# Create a simple position argument
pos = [1.0, 2.0, 3.0]

ARGS = {
    'pos': pos
}
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 dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict 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
unit_vector Ran
ran, violated the contract we drafted
vinnik-dmitry07/minimal-hand/app1.py
code served (permissive licence) · get_code("b43e0bf3e224c756")
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 1, dtype family float, shape (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

def unit_vector(vector):
    return vector / np.linalg.norm(vector)

# Create a small numpy vector
vector = np.array([3.0, 4.0, 0.0])

ARGS = {
    "vector": vector
}
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 1, dtype family float, shape (3,) — shape (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
xyz_to_delta Ran
ran on a synthesized input
CalciferZh/minimal-hand/kinematics.py
code served (permissive licence) · get_code("06f6fcc9a30c5f4e")
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

class JointsDef:
    """Stand-in class for joints definition like MPIIHandJoints."""
    def __init__(self, n_joints, parents):
        self.n_joints = n_joints
        self.parents = parents

# Create a simple skeleton: 4 joints
# Joint 0 is root (parent=None)
# Joint 1's parent is joint 0
# Joint 2's parent is joint 1
# Joint 3's parent is joint 0
joints_def = JointsDef(
    n_joints=4,
    parents=[None, 0, 1, 0]
)

# Create joint coordinates: 4 joints, 3D coordinates
xyz = np.array([
    [0.0, 0.0, 0.0],   # joint 0 (root)
    [1.0, 0.0, 0.0],   # joint 1
    [2.0, 1.0, 0.0],   # joint 2
    [0.0, 1.0, 0.0],   # joint 3
], dtype=np.float32)

ARGS = {
    "xyz": xyz,
    "joints_def": joints_def,
}
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 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
zero_padding Ran
ran on a synthesized input
CalciferZh/minimal-hand/network.py
code served (permissive licence) · get_code("cb6951cec018a217")
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 EagerTensor, ndim 4, dtype family float, shape (2, 7, 7, 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 tensorflow as tf

# Create a small TensorFlow tensor to use as input
# The function expects a 4D tensor (batch, height, width, channels)
inputs = tf.constant(np.random.randn(2, 4, 4, 3), dtype=tf.float32)

# Define padding amounts for dimensions 1 and 2
pad_1 = 1
pad_2 = 2

ARGS = {
    "inputs": inputs,
    "pad_1": pad_1,
    "pad_2": pad_2,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, tensorflow — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type EagerTensor, ndim 4, dtype family float, shape (2, 7, 7, 3) — shape (2, 7, 7, 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
conv_bn Not yet run
unverified (harvested, no recorded run)
CalciferZh/minimal-hand/network.py
code served (permissive licence) · get_code("d62dd73c3d44c857")
How it ran

no invocation recorded.

conv_bn Not yet run
unverified (harvested, no recorded run)
vinnik-dmitry07/minimal-hand/network.py
code served (permissive licence) · get_code("25237f9d8ced13f2")
How it ran

no invocation recorded.

conv_bn_relu Not yet run
unverified (harvested, no recorded run)
CalciferZh/minimal-hand/network.py
code served (permissive licence) · get_code("4d2a66359100655a")
How it ran

no invocation recorded.

conv_bn_relu Not yet run
unverified (harvested, no recorded run)
vinnik-dmitry07/minimal-hand/network.py
code served (permissive licence) · get_code("958f62a80012eb81")
How it ran

no invocation recorded.

Also present in other papers' code

Of 14 samples lifted from this paper's linked repositories, 14 carry a function identity and 0 do not. Of the 13 distinct functions with an identity, 2 are also present in other papers whose code we hold (1 with at least one paper dated earlier, 0 the same month, 2 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.

FunctionAlso present inEarliest-datedOthers named
unit_vector 4 other papers
1 dated earlier, 3 later
1904.12659 (2019-04)
Actional-Structural Graph Convolutional Networks for Skeleton-based Action Recognition
2104.00946 (2021-04)
UAV-Human: A Large Benchmark for Human Behavior Understanding with Unmanned Aerial Vehicles
aaai_25127 (2023 (year only))
Hierarchical Contrast for Unsupervised Skeleton-Based Action Representation Learning
and 1 more not named here
load_pkl 2 other papers
2 later
2203.02167 (2022-03)
SimKGC: Simple Contrastive Knowledge Graph Completion with Pre-trained Language Models
2508.17554 (2025-08)
Bridging Graph and State-Space Modeling for Intensive Care Unit Length of Stay Prediction

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 a novel method for monocular hand shape and pose estimation at unprecedented runtime performance of 100fps and at state-of-the-art accuracy. This is enabled by a new learning based architecture designed such that it can make use of all the sources of available hand training data: image data with either 2D or 3D annotations, as well as stand-alone 3D animations without corresponding image data. It features a 3D hand joint detection module and an inverse kinematics module which regresses not only 3D joint positions but also maps them to joint rotations in a single feed-forward pass. This output makes the method more directly usable for applications in computer vision and graphics compared to only regressing 3D joint positions. We demonstrate that our architectural design leads to a significant quantitative and qualitative improvement over the state of the art on several challenging benchmarks. Our model is publicly available for future research.

For agents

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

get_harvested_code_for_paper("2003.09572")
get_code_for_paper("2003.09572")
have("2003.09572")

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

curl https://syntology.ai/api/ran/2003.09572.json

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

[![Syntology run record](https://syntology.ai/api/ran/2003.09572.svg)](https://syntology.ai/paper/2003.09572)

Connect an agent — have() is free.