SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2008.00188 · 2020

Augmented Skeleton Based Contrastive Action Learning with Momentum LSTM for Unsupervised Action Recognition

arXiv · PDF · Open in the Atlas

Code that ran

9 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 7 have no contract verdict; 2 failures of our instrument; 2 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/2008.00188.json · badge: /api/ran/2008.00188.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
Mikexu007/AS_CAL canonical (classifier label)
classifier basis: census
9 ran · 2 no recorded run
FunctionStatusWhere it lives
denormalize Ran
ran on a synthesized input
Mikexu007/AS_CAL/feeders/sbu_feeder.py
code served (permissive licence) · get_code("dc89adec7492521a")
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 (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 numpy as np

# Create a small normalized coordinates array with shape (2, 3)
# The function expects normalized coordinates where:
# - Column 0: normalized_X (will be denormalized using formula: 1280 - X * 2560)
# - Column 1: normalized_Y (will be denormalized using formula: 960 - Y * 1920)
# - Column 2: normalized_Z (will be denormalized using formula: Z * 10000 / 7.8125)

norm_coords = np.array([
    [0.1, 0.2, 0.3],
    [0.4, 0.5, 0.6]
], dtype=np.float32)

ARGS = {
    "norm_coords": norm_coords
}
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 (2, 3) — shape (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
get_ground_truth Ran
ran on a synthesized input
Mikexu007/AS_CAL/feeders/sbu_feeder.py
code served (permissive licence) · get_code("63dca04df5440f53")
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
import pandas as pd
import os
import tempfile

# Create a temporary directory structure for testing
temp_dir = tempfile.mkdtemp()

# Create mock skeleton files
SETS = ['s01s02', 's01s03', 's01s07', 's02s01', 's02s03', 's02s06', 's02s07', 's03s02', 's03s04', 's03s05', 's03s06', 's04s02', 's04s03', 's04s06', 's05s02', 's05s03', 's06s02', 's06s03', 's06s04', 's07s01', 's07s03']
FOLDS = [[1, 9, 15, 19], [5, 7, 10, 16], [2, 3, 20, 21], [4, 6, 8, 11], [12, 13, 14, 17, 18]]

# Create a few test directories with skeleton files
for set_idx, set_name in enumerate(SETS[:2]):  # Just use first 2 sets
    for action_id in range(2):  # Just 2 actions
        for seq_id in range(1):  # Just 1 sequence per action
            dir_path = os.path.join(temp_dir, set_name, f'{action_id + 1:02d}', f'seq_{seq_id}')
            os.makedirs(dir_path, exist_ok=True)
            
            # Create a mock skeleton_pos.txt file with 90 columns (2 people * 45 joints)
            # Create 5 frames
            skeleton_data = np.random.rand(5, 90)
            filepath = os.path.join(dir_path, 'skeleton_pos.txt')
            
            # Add frame index as first column, then 90 columns
            frame_indices = np.arange(5).reshape(-1, 1)
            full_data = np.hstack([frame_indices, skeleton_data])
            
            np.savetxt(filepath, full_data, delimiter=',', fmt='%.6f')

ARGS = {
    'data_dir': temp_dir
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, pandas — 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
get_shuffle_ids Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
Mikexu007/AS_CAL/pretrain_and_linEval.py
code served (permissive licence) · get_code("e2e91e39d8368ced")
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) and returned tuple, ndim 2, dtype family int; no output shape is stated. The image it ran in was not recorded. The code mentions cuda, and the harness aliases cuda to the CPU for every sample; whether this run needed that is 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)

# The function get_shuffle_ids takes a batch size (bsz) as input
# and generates shuffle indices for ShuffleBN operations.
# It requires an integer batch size parameter.

bsz = 2

ARGS = {
    "bsz": bsz
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
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 2, dtype family int — 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
import_class Ran
ran; our contract draft was wrong, not the code
Mikexu007/AS_CAL/linEval.py
pointer only (licence: MIT); the stored copy get_code serves, from iamjeff7/j-va-aagcn, is pointer only (licence: NONE) · get_code("ffed4f85d50832c9")
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 iamjeff7/j-va-aagcn, and GitHub detected no licence in that repository 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.

parse_sbu_txt Ran
ran on a synthesized input
Mikexu007/AS_CAL/feeders/sbu_feeder.py
code served (permissive licence) · get_code("5deb486c8215ef31")
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 4, dtype family float, shape (3, 2, 15, 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 tempfile
import os

# Create a temporary CSV file with the expected format
# The function uses np.loadtxt with usecols=range(1, 91), so we need at least 91 columns
# Each row should have 90 values (columns 1-90, skipping column 0)
# The function expects 2 people * 45 values per person = 90 values per frame

temp_file = tempfile.NamedTemporaryFile(mode='w', delete=False, suffix='.txt')
pose_filepath = temp_file.name

# Create sample data: 3 frames with 91 columns (column 0 is skipped, columns 1-90 are used)
# Each frame has 90 values representing 2 people with 15 joints * 3 coordinates each
np.random.seed(42)
data = np.random.rand(3, 91)
np.savetxt(temp_file, data, delimiter=',', fmt='%.6f')
temp_file.close()

ARGS = {
    'pose_filepath': pose_filepath,
    'normalized': False
}
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 4, dtype family float, shape (3, 2, 15, 3) — shape (3, 2, 15, 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
pose_embed_ Ran
ran on a synthesized input
Mikexu007/AS_CAL/linEval.py
code served (permissive licence) · get_code("16950c872dc57466")
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 (5, 2, 64) 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)

pose_embedding_size = 64

class PoseModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(4, pose_embedding_size)
    
    def forward(self, x):
        # x shape: (batch, 4)
        # returns shape: (batch, pose_embedding_size)
        return self.linear(x)

# Test case 1: 3D input (t, n, d)
x_3d = torch.randn(5, 2, 4)  # t=5, n=2, d=4
t_val = 5
n_val = 2
pose_model = PoseModel()

ARGS = {
    "x": x_3d,
    "t": t_val,
    "n": n_val,
    "pose_model": pose_model
}
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 (5, 2, 64) — shape (5, 2, 64) 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
reshap_input_for_lstm Ran
ran on a synthesized input
Mikexu007/AS_CAL/models/Models.py
code served (permissive licence) · get_code("d82afa8773b11f30")
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 (4, 2, 90) 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)

# The function expects a 5D tensor with dimensions (n, c, t, v, m)
# n = batch size
# c = channels
# t = time steps
# v = vertices/spatial dimension
# m = number of people/bodies

# Create a small 5D tensor with distinct sizes for each dimension
x1 = torch.randn(2, 3, 4, 5, 6)  # (n=2, c=3, t=4, v=5, m=6)

ARGS = {
    "x1": x1
}
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 (4, 2, 90) — shape (4, 2, 90) 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
subsample Ran
ran on a synthesized input
Mikexu007/AS_CAL/feeders/tools.py
code served (permissive licence) · get_code("d4cef4738b78026c")
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 4, dtype family float, shape (3, 10, 5, 2) 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 random

# Set seed for reproducibility
random.seed(0)
np.random.seed(0)

# Create sample data
# Shape: (C, T, V, M) = (Channels, Time, Vertices, Persons)
C, T, V, M = 3, 10, 5, 2
data_numpy = np.random.randn(C, T, V, M).astype(np.float32)

# time_range can be either an int or a list
# Testing with an int - will randomly sample that many frames
time_range = 5

ARGS = {
    "data_numpy": data_numpy,
    "time_range": time_range,
}
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 4, dtype family float, shape (3, 10, 5, 2) — shape (3, 10, 5, 2) 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
subtract Ran
ran on a synthesized input
Mikexu007/AS_CAL/feeders/tools.py
code served (permissive licence) · get_code("5772bc1d78b90f56")
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 4, dtype family float, shape (3, 4, 5, 2) 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

# Create a small numpy array with shape (C, T, V, M)
# Using distinct sizes: C=3, T=4, V=5, M=2
C, T, V, M = 3, 4, 5, 2
data_numpy = np.random.randn(C, T, V, M).astype(np.float32)

# joint parameter should be a valid index into the V dimension
joint = 1

ARGS = {
    "data_numpy": data_numpy,
    "joint": joint,
}
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 4, dtype family float, shape (3, 4, 5, 2) — shape (3, 4, 5, 2) 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
accuracy Not yet run
unverified (harvested, no recorded run)
Mikexu007/AS_CAL/util.py
pointer only (licence: MIT); the stored copy get_code serves, from boone891214/MEST, is pointer only (licence: NONE) · get_code("913d82af53065418")
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 boone891214/MEST, and GitHub detected no licence in that repository 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. No invocation recorded.

aug_look Not yet run
unverified (harvested, no recorded run)
Mikexu007/AS_CAL/feeders/tools.py
code served (permissive licence) · get_code("fec3ca8e9cb588e8")
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, 3 are also present in other papers whose code we hold (3 with at least one paper dated earlier, 1 the same month, 3 later) and 8 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
accuracy 80 other papers
43 dated earlier, 2 the same month, 35 later
1307.0414 (2013-07)
Challenges in Representation Learning: A report on three machine learning contests
1412.6553 (2014-12)
Speeding-up Convolutional Neural Networks Using Fine-tuned CP-Decomposition
1503.02531 (2015-03)
Distilling the Knowledge in a Neural Network
and 77 more not named here
import_class 18 other papers
2 dated earlier, 16 later · 1 only as an assembly brick
1804.07453 (2018-04)
View Adaptive Neural Networks for High Performance Skeleton-based Human Action Recognition
1911.04131 (2019-11)
Learning Graph Convolutional Network for Skeleton-based Human Action Recognition by Neural Searching
2011.13322 (2020-11)
Spatio-Temporal Inception Graph Convolutional Networks for Skeleton-Based Action Recognition
and 15 more not named here
get_shuffle_ids 9 other papers
2 dated earlier, 7 later · 1 only as an assembly brick
1911.05722 (2019-11)
Momentum Contrast for Unsupervised Visual Representation Learning
2007.07423 (2020-07)
Comparing to Learn: Surpassing ImageNet Pretraining on Radiographs By Comparing Image Representations
2105.07269 (2021-05)
Mean Shift for Self-Supervised Learning
and 6 more not named here

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

Action recognition via 3D skeleton data is an emerging important topic in these years. Most existing methods either extract hand-crafted descriptors or learn action representations by supervised learning paradigms that require massive labeled data. In this paper, we for the first time propose a contrastive action learning paradigm named AS-CAL that can leverage different augmentations of unlabeled skeleton data to learn action representations in an unsupervised manner. Specifically, we first propose to contrast similarity between augmented instances (query and key) of the input skeleton sequence, which are transformed by multiple novel augmentation strategies, to learn inherent action patterns ("pattern-invariance") of different skeleton transformations. Second, to encourage learning the pattern-invariance with more consistent action representations, we propose a momentum LSTM, which is implemented as the momentum-based moving average of LSTM based query encoder, to encode long-term action dynamics of the key sequence. Third, we introduce a queue to store the encoded keys, which allows our model to flexibly reuse proceeding keys and build a more consistent dictionary to improve contrastive learning. Last, by temporally averaging the hidden states of action learned by the query encoder, a novel representation named Contrastive Action Encoding (CAE) is proposed to represent human's action effectively. Extensive experiments show that our approach typically improves existing hand-crafted methods by 10-50% top-1 accuracy, and it can achieve comparable or even superior performance to numerous supervised learning methods.

For agents

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

get_harvested_code_for_paper("2008.00188")
get_code_for_paper("2008.00188")
have("2008.00188")

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

curl https://syntology.ai/api/ran/2008.00188.json

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

[![Syntology run record](https://syntology.ai/api/ran/2008.00188.svg)](https://syntology.ai/paper/2008.00188)

Connect an agent — have() is free.