SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2006.10214 · 2020

MediaPipe Hands: On-device Real-time Hand Tracking

arXiv · PDF · Open in the Atlas

Code that ran

6 of 7 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 4 have no contract verdict; 2 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/2006.10214.json · badge: /api/ran/2006.10214.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
vidursatija/BlazePalm community (classifier label)
classifier basis not recorded on the link
3 ran · 1 no recorded run
Matheus-Vyctor/Gesture_Hand_Controller-1 community (classifier label)
classifier basis not recorded on the link
3 ran · 0 no recorded run
FunctionStatusWhere it lives
calculate_scale Ran
ran on a synthesized input
vidursatija/BlazePalm/ML/genarchors.py
code served (permissive licence) · get_code("dc684f7292e42a30")
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)

# The function calculate_scale takes numeric parameters and performs a linear interpolation
# min_scale: lower bound of the scale
# max_scale: upper bound of the scale
# stride_index: current index in the stride sequence
# num_strides: total number of strides

ARGS = {
    "min_scale": 0.2,
    "max_scale": 0.9,
    "stride_index": 2,
    "num_strides": 5
}
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
get_angle Ran
ran on a synthesized input
Matheus-Vyctor/Gesture_Hand_Controller-1/utils.py
code served (permissive licence) · get_code("58e20fa0b56397c9")
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 float32, 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 get_angle takes hand_landmarks as input
# hand_landmarks is used in get_points_to_zoom which indexes it as:
# lm[24], lm[25], lm[12], lm[13], lm[3], lm[4]
# So we need at least 26 elements (indices 0-25)

# Create a hand_landmarks array with at least 26 elements
# Each element should be a coordinate value (float)
hand_landmarks = np.array([
    0.1, 0.2, 0.3, 0.4,  # indices 0-3
    0.5, 0.6, 0.7, 0.8,  # indices 4-7
    0.9, 1.0, 1.1, 1.2,  # indices 8-11
    1.3, 1.4, 1.5, 1.6,  # indices 12-15
    1.7, 1.8, 1.9, 2.0,  # indices 16-19
    2.1, 2.2, 2.3, 2.4,  # indices 20-23
    2.5, 2.6              # indices 24-25
], dtype=np.float32)

ARGS = {
    "hand_landmarks": hand_landmarks
}
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 float32, 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
get_average_points Ran
ran on a synthesized input
Matheus-Vyctor/Gesture_Hand_Controller-1/utils.py
code served (permissive licence) · get_code("3eef38e028e06d95")
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 1, 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
import numpy as np

torch.manual_seed(0)

# The function accesses lm[12], lm[24], lm[36], lm[48], lm[60] for x coordinates
# and lm[13], lm[25], lm[37], lm[49], lm[61] for y coordinates
# So lm needs to be indexable with indices up to 61, with at least 62 elements
# Each element should be a numeric value (x or y coordinate)

# Create a landmark array with 62 elements (indices 0-61)
# This appears to be facial landmark points, so we'll use reasonable coordinate values
lm = np.random.rand(62) * 100  # Random coordinates between 0-100

ARGS = {
    "lm": lm
}
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, ndim 1, 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
get_points_to_zoom Ran
ran on a synthesized input
Matheus-Vyctor/Gesture_Hand_Controller-1/utils.py
code served (permissive licence) · get_code("3376da5e2befda7c")
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 2, 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 numpy as np

# The function expects `lm` to be array-like with at least 26 elements
# (indices 0-25 are accessed, specifically indices 3, 4, 12, 13, 24, 25)
# Create a landmark array with sufficient size
lm = np.array([
    0.1, 0.2,  # 0, 1
    0.3, 0.4,  # 2, 3
    0.5, 0.6,  # 4, 5
    0.7, 0.8,  # 6, 7
    0.9, 1.0,  # 8, 9
    1.1, 1.2,  # 10, 11
    1.3, 1.4,  # 12, 13 - thumb_tip
    1.5, 1.6,  # 14, 15
    1.7, 1.8,  # 16, 17
    1.9, 2.0,  # 18, 19
    2.1, 2.2,  # 20, 21
    2.3, 2.4,  # 22, 23
    2.5, 2.6,  # 24, 25 - index_finger_tip
])

ARGS = {
    "lm": lm
}
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, ndim 2, 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
intersect Ran
ran; our contract draft was wrong, not the code
vidursatija/BlazePalm/ML/blazepalm.py
pointer only (licence: MIT); the stored copy get_code serves, from NVlabs/AL-MDN, is pointer only (licence: NOASSERTION) · get_code("f01c5f73515770a5")
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 NVlabs/AL-MDN, and GitHub could not classify the licence of that repository (NOASSERTION). 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.

jaccard Ran
ran; our contract draft was wrong, not the code
vidursatija/BlazePalm/ML/blazepalm.py
pointer only (licence: MIT); the stored copy get_code serves, from NVlabs/AL-MDN, is pointer only (licence: NOASSERTION) · get_code("86489779d4c76926")
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 NVlabs/AL-MDN, and GitHub could not classify the licence of that repository (NOASSERTION). 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.

overlap_similarity Not yet run
unverified (harvested, no recorded run)
vidursatija/BlazePalm/ML/blazepalm.py
code served (permissive licence) · get_code("a0ca886f22744fea")
How it ran

no invocation recorded.

Also present in other papers' code

Of 7 samples lifted from this paper's linked repositories, 7 carry a function identity and 0 do not. Of the 7 distinct functions with an identity, 3 are also present in other papers whose code we hold (3 with at least one paper dated earlier, 1 the same month, 2 later) and 4 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
intersect 14 other papers
8 dated earlier, 1 the same month, 5 later
1409.1556 (2014-09)
Very Deep Convolutional Networks for Large-Scale Image Recognition
1512.02325 (2015-12)
SSD: Single Shot MultiBox Detector
and 11 more not named here
jaccard 2 other papers
1 dated earlier, 1 later
1703.07332 (2017-03)
How far are we from solving the 2D & 3D Face Alignment problem? (and a dataset of 230,000 3D facial landmarks)
2103.16130 (2021-03)
Active Learning for Deep Object Detection via Probabilistic Modeling
overlap_similarity 1 other paper
1 dated earlier
1703.07332 (2017-03)
How far are we from solving the 2D & 3D Face Alignment problem? (and a dataset of 230,000 3D facial landmarks)

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 real-time on-device hand tracking pipeline that predicts hand skeleton from single RGB camera for AR/VR applications. The pipeline consists of two models: 1) a palm detector, 2) a hand landmark model. It's implemented via MediaPipe, a framework for building cross-platform ML solutions. The proposed model and pipeline architecture demonstrates real-time inference speed on mobile GPUs and high prediction quality. MediaPipe Hands is open sourced at https://mediapipe.dev.

For agents

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

get_harvested_code_for_paper("2006.10214")
get_code_for_paper("2006.10214")
have("2006.10214")

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

curl https://syntology.ai/api/ran/2006.10214.json

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

[![Syntology run record](https://syntology.ai/api/ran/2006.10214.svg)](https://syntology.ai/paper/2006.10214)

Connect an agent — have() is free.