SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 136820244 · ECCV · 2022

Diverse Human Motion Prediction Guided by Multi-Level Spatial-Temporal Anchors

Yu-Xiong Wang, Sirui Xu, Liang-Yan Gui

Open in the Atlas

Code that ran

6 of 8 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; 2 honoured the contract we drafted, 1 violated it and 3 have no contract verdict; 0 failures of our instrument; 2 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/136820244.json · badge: /api/ran/136820244.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
Sirui-Xu/STARS canonical (classifier label)
classifier basis: census
6 ran · 2 no recorded run
FunctionStatusWhere it lives
final_mpjpe_error Ran
ran on a synthesized input
Sirui-Xu/STARS/deterministic/utils/loss_funcs.py
code served (permissive licence) · get_code("e4e64f856da66dff")
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 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)

# Analyzing the function:
# - batch_pred: indexed with [:, -1, :, :] then viewed as (-1, 3)
#   This means it has shape (batch, time, joints, 3)
# - batch_gt: same indexing pattern, same shape
# - The function computes per-joint 3D distances and takes the mean

# Using batch=2, time=4, joints=5 (distinct sizes for different dimensions)
batch_size = 2
time_steps = 4
num_joints = 5
coords = 3

batch_pred = torch.randn(batch_size, time_steps, num_joints, coords)
batch_gt = torch.randn(batch_size, time_steps, num_joints, coords)

ARGS = {
    "batch_pred": batch_pred,
    "batch_gt": batch_gt,
}
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 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
mpjpe_error Ran
ran on a synthesized input
Sirui-Xu/STARS/deterministic/utils/loss_funcs.py
code served (permissive licence) · get_code("a34f149c8de491d2")
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 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)

# Create small tensors for batch_pred and batch_gt
# The function reshapes them to (-1, 3), so we need tensors that can be reshaped that way
# Using shape (2, 4, 3) for predicted batch and (2, 4, 3) for ground truth
# This represents 2 samples, 4 joints each, 3 coordinates (x, y, z)

batch_pred = torch.randn(2, 4, 3)
batch_gt = torch.randn(2, 4, 3)

ARGS = {
    "batch_pred": batch_pred,
    "batch_gt": batch_gt,
}
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 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
rotmat2euler Ran
ran, honoured the contract we drafted
Sirui-Xu/STARS/deterministic/utils/data_utils.py
code served (permissive licence) · get_code("7e2daa491ccfd852")
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

# Construct a valid 3x3 rotation matrix
# Using a simple rotation around the z-axis
angle = np.pi / 4
R = np.array([
    [np.cos(angle), -np.sin(angle), 0],
    [np.sin(angle), np.cos(angle), 0],
    [0, 0, 1]
])

ARGS = {
    "R": R
}
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
rotmat2expmap Ran
ran, violated the contract we drafted
Sirui-Xu/STARS/deterministic/utils/data_utils.py
code served (permissive licence) · get_code("2796e9930642e3d3")
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

# Create a valid 3x3 rotation matrix
# Using a simple rotation matrix around z-axis by 45 degrees
angle = np.pi / 4
R = np.array([
    [np.cos(angle), -np.sin(angle), 0],
    [np.sin(angle), np.cos(angle), 0],
    [0, 0, 1]
], dtype=np.float32)

ARGS = {
    "R": R
}
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
rotmat2quat Ran
ran, honoured the contract we drafted
Sirui-Xu/STARS/deterministic/utils/data_utils.py
code served (permissive licence) · get_code("bf9baa714c29fd46")
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 (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 numpy as np

# Create a valid 3x3 rotation matrix
# Using a simple rotation matrix around the z-axis
angle = np.pi / 4
R = np.array([
    [np.cos(angle), -np.sin(angle), 0],
    [np.sin(angle), np.cos(angle), 0],
    [0, 0, 1]
], dtype=np.float32)

ARGS = {
    "R": R
}
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 (4,) — shape (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
update Ran
ran on a synthesized input
Sirui-Xu/STARS/deterministic/utils/h36_3d_viz.py
code served (permissive licence) · get_code("98c8a875f6a616cf")
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 numpy as np
import matplotlib.pyplot as plt
from mpl_toolkits.mplot3d import Axes3D

# Create a small 3D plot figure and axis
fig = plt.figure()
ax = fig.add_subplot(111, projection='3d')

# Create small pose data (32 joints, 3D coordinates)
# Shape: (num_frames, num_joints, 3)
num_joints = 32
num_frames = 10
input_n = 5

# Ground truth data: 10 frames of 32 joints with (x, y, z) coordinates
data_gt = np.random.randn(num_frames, num_joints, 3).astype(np.float32)

# Predicted data: same shape
data_pred = np.random.randn(num_frames, num_joints, 3).astype(np.float32)

# Initialize empty plot lists (will be populated by create_pose)
plots_gt = []
plots_pred = []

# Pre-populate plots by calling create_pose for frame 0
plots_gt = []
plots_pred = []

# Create initial plots for both ground truth and prediction
gt_vals = data_gt[0]
pred_vals = data_pred[0]

# Manually create the plot objects (simulating what create_pose does)
connect = [(1, 2), (2, 3), (3, 4), (4, 5), (6, 7), (7, 8), (8, 9), (9, 10), (0, 1), (0, 6), (6, 17), (17, 18), (18, 19), (19, 20), (20, 21), (21, 22), (1, 25), (25, 26), (26, 27), (27, 28), (28, 29), (29, 30), (24, 25), (24, 17), (24, 14), (14, 15)]

for i in np.arange(len(connect)):
    x = np.array([gt_vals[connect[i][0], 0], gt_vals[connect[i][1], 0]])
    y = np.array([gt_vals[connect[i][0], 1], gt_vals[connect[i][1], 1]])
    z = np.array([gt_vals[connect[i][0], 2], gt_vals[connect[i][1], 2]])
    plots_gt.append(ax.plot(x, y, z, lw=2, linestyle='--', c='#8e8e8e'))
    
    x = np.array([pred_vals[connect[i][0], 0], pred_vals[connect[i][1], 0]])
    y = np.array([pred_vals[connect[i][0], 1], pred_vals[connect[i][1], 1]])
    z = np.array([pred_vals[connect[i][0], 2], pred_vals[connect[i][1], 2]])
    plots_pred.append(ax.plot(x, y, z, lw=2, linestyle='--', c='#9b59b6'))

ARGS = {
    'num': 1,
    'data_gt': data_gt,
    'data_pred': data_pred,
    'plots_gt': plots_gt,
    'plots_pred': plots_pred,
    'fig': fig,
    'ax': ax,
    'input_n': input_n
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
matplotlib, mpl_toolkits, numpy — 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
create_pose Not yet run
unverified (harvested, no recorded run)
Sirui-Xu/STARS/deterministic/utils/h36_3d_viz.py
code served (permissive licence) · get_code("de1e9376e6a341ee")
How it ran

no invocation recorded.

loss_function Not yet run
unverified (harvested, no recorded run)
Sirui-Xu/STARS/diverse/train_nf.py
code served (permissive licence) · get_code("3671a5b389b22a4d")
How it ran

no invocation recorded.

Also present in other papers' code

Of 8 samples lifted from this paper's linked repositories, 8 carry a function identity and 0 do not. Of the 8 distinct functions with an identity, 4 are also present in other papers whose code we hold (4 with at least one paper dated earlier, 0 the same month, 3 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
rotmat2euler 4 other papers
3 dated earlier, 1 later
1805.00655 (2018-05)
Convolutional Sequence to Sequence Model for Human Dynamics
1809.03036 (2018-09)
A Neural Temporal Model for Human Motion Prediction
1908.05436 (2019-08)
Learning Trajectory Dependencies for Human Motion Prediction
and 1 more not named here
rotmat2quat 4 other papers
3 dated earlier, 1 later
1805.00655 (2018-05)
Convolutional Sequence to Sequence Model for Human Dynamics
1809.03036 (2018-09)
A Neural Temporal Model for Human Motion Prediction
1908.05436 (2019-08)
Learning Trajectory Dependencies for Human Motion Prediction
and 1 more not named here
rotmat2expmap 2 other papers
1 dated earlier, 1 later
1908.05436 (2019-08)
Learning Trajectory Dependencies for Human Motion Prediction
2308.08942 (2023-08)
Auxiliary Tasks Benefit 3D Skeleton-based Human Motion Prediction
loss_function 1 other paper
1 dated earlier
2108.08422 (2021-08)
Generating Smooth Pose Sequences for Diverse Human Motion 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

Predicting diverse human motions given a sequence of historical poses has received increasing attention. Despite rapid progress, existing work captures the multi-modal nature of human motions primarily through likelihood-based sampling, where the mode collapse has been widely observed. In this paper, we propose a simple yet effective approach that disentangles randomly sampled codes with a deterministic learnable component named anchors to promote sample precision and diversity. Anchors are further factorized into spatial anchors and temporal anchors, which provide attractively interpretable control over spatialtemporal disparity. In principle, our spatial-temporal anchor-based sampling (STARS) can be applied to different motion predictors. Here we propose an interaction-enhanced spatial-temporal graph convolutional network (IE-STGCN) that encodes prior knowledge of human motions (e.g., spatial locality), and incorporate the anchors into it. Extensive experiments demonstrate that our approach outperforms state of the art in both stochastic and deterministic prediction, suggesting it as a unified framework for modeling human motions. Our code and pretrained models are available at https://github.com/Sirui-Xu/STARS.

For agents

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

get_harvested_code_for_paper("136820244")
get_code_for_paper("136820244")
have("136820244")

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

curl https://syntology.ai/api/ran/136820244.json

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

[![Syntology run record](https://syntology.ai/api/ran/136820244.svg)](https://syntology.ai/paper/136820244)

Connect an agent — have() is free.