SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2204.12484 · NeurIPS · 2022

ViTPose: Simple Vision Transformer Baselines for Human Pose Estimation

Dacheng Tao, Jing Zhang, Yufei Xu, Qiming Zhang

arXiv · PDF · Open in the Atlas

Code that ran

18 of 31 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (8 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 10 have no contract verdict; 8 failures of our instrument; 13 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/2204.12484.json · badge: /api/ran/2204.12484.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
JunkyByte/easy_ViTPose application (classifier label)
classifier basis: classifier_2025
12 ran · 8 no recorded run
gpastal24/ViTPose-Pytorch port (classifier label)
classifier basis: classifier_2025
5 ran · 1 no recorded run
jaehyunnn/ViTPose_pytorch reimplementation (classifier label)
classifier basis: classifier_2025
1 ran · 4 no recorded run
FunctionStatusWhere it lives
Attention Ran
ran on a synthesized input
gpastal24/ViTPose-Pytorch/src/vitpose_infer/builder/backbones/vit.py
pointer only (licence: GPL-3.0) · get_code("a17528cd8209371a")
How it ran

The body of this sample is withheld: GPL-3.0, as GitHub detects it on this paper's repository, gpastal24/ViTPose-Pytorch, is not in the permissive set Syntology serves inline. 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.

Block Ran
ran on a synthesized input · constructed an object; no result was computed
gpastal24/ViTPose-Pytorch/src/vitpose_infer/builder/backbones/vit.py
pointer only (licence: GPL-3.0) · get_code("daeb52387dc365ad")
How it ran

The body of this sample is withheld: GPL-3.0, as GitHub detects it on this paper's repository, gpastal24/ViTPose-Pytorch, is not in the permissive set Syntology serves inline. 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.

Block Ran
ran on a synthesized input · constructed an object; no result was computed
JunkyByte/easy_ViTPose/easy_ViTPose/vit_models/model.py
code served (permissive licence) · get_code("b45be27e9755bcea")
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 an instance of its own class, Block: the run constructed the object and did not call it, so no result was computed. 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 torch.nn as nn

torch.manual_seed(0)

ARGS = {
    "dim": 64,
    "num_heads": 8,
    "mlp_ratio": 4.0,
    "qkv_bias": False,
    "qk_scale": None,
    "drop": 0.1,
    "attn_drop": 0.1,
    "drop_path": 0.1,
    "act_layer": nn.GELU,
    "norm_layer": nn.LayerNorm,
    "attn_head_dim": None,
}
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 Block — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a Block 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
DropPath Ran
ran on a synthesized input · constructed an object; no result was computed
gpastal24/ViTPose-Pytorch/src/vitpose_infer/builder/backbones/vit.py
pointer only (licence: GPL-3.0) · get_code("cd0e7be073ad785f")
How it ran

The body of this sample is withheld: GPL-3.0, as GitHub detects it on this paper's repository, gpastal24/ViTPose-Pytorch, is not in the permissive set Syntology serves inline. 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.

DropPath Ran
ran on a synthesized input · constructed an object; no result was computed
JunkyByte/easy_ViTPose/easy_ViTPose/vit_models/model.py
code served (permissive licence) · get_code("bcec6a4434dcfdb8")
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 an instance of its own class, DropPath: the run constructed the object and did not call it, so no result was computed. 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 torch.nn as nn

torch.manual_seed(0)

# Create a small input tensor for testing
# Shape: (batch_size=2, channels=4, height=3, width=3)
x = torch.randn(2, 4, 3, 3)

# DropPath.__init__ only takes drop_prob parameter
ARGS = {
    'drop_prob': 0.5
}
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 DropPath — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a DropPath 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
HybridEmbed Ran
ran on a synthesized input · constructed an object; no result was computed
gpastal24/ViTPose-Pytorch/src/vitpose_infer/builder/backbones/vit.py
pointer only (licence: GPL-3.0) · get_code("6f5efea0780d30f0")
How it ran

The body of this sample is withheld: GPL-3.0, as GitHub detects it on this paper's repository, gpastal24/ViTPose-Pytorch, is not in the permissive set Syntology serves inline. 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.

HybridEmbed Ran
ran on a synthesized input · constructed an object; no result was computed
JunkyByte/easy_ViTPose/easy_ViTPose/vit_models/model.py
code served (permissive licence) · get_code("3729451b09b4c285")
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 an instance of its own class, HybridEmbed: the run constructed the object and did not call it, so no result was computed. 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 torch.nn as nn

torch.manual_seed(0)

# Create a minimal backbone that satisfies HybridEmbed's requirements
class MinimalBackbone(nn.Module):
    def __init__(self, in_chans=3, feature_dim=64):
        super().__init__()
        self.feature_dim = feature_dim
        self.in_chans = in_chans
        # Simple conv layer to process input
        self.conv = nn.Conv2d(in_chans, feature_dim, kernel_size=3, padding=1)
        self.feature_info = type('obj', (object,), {'channels': lambda: [feature_dim]})()
    
    def forward(self, x):
        # x shape: (batch, in_chans, H, W)
        # Output should be a list/tuple where last element is feature map
        # Feature map shape should be (batch, feature_dim, H', W')
        x = self.conv(x)
        return [x]  # Return as list so [-1] indexing works

# Create backbone instance
backbone = MinimalBackbone(in_chans=3, feature_dim=64)

# Define ARGS dict with all parameters for HybridEmbed
ARGS = {
    'backbone': backbone,
    'img_size': 224,
    'feature_size': None,
    'in_chans': 3,
    'embed_dim': 768,
}
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 HybridEmbed — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a HybridEmbed 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
PatchEmbed Ran
ran on a synthesized input · constructed an object; no result was computed
gpastal24/ViTPose-Pytorch/src/vitpose_infer/builder/backbones/vit.py
pointer only (licence: GPL-3.0) · get_code("fa78df3f00a52ac8")
How it ran

The body of this sample is withheld: GPL-3.0, as GitHub detects it on this paper's repository, gpastal24/ViTPose-Pytorch, is not in the permissive set Syntology serves inline. 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.

PatchEmbed Ran
ran on a synthesized input · constructed an object; no result was computed
JunkyByte/easy_ViTPose/easy_ViTPose/vit_models/model.py
code served (permissive licence) · get_code("38811a374b1cff76")
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 an instance of its own class, PatchEmbed: the run constructed the object and did not call it, so no result was computed. 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 torch.nn as nn

torch.manual_seed(0)

ARGS = {
    'img_size': 224,
    'patch_size': 16,
    'in_chans': 3,
    'embed_dim': 768,
    'ratio': 1
}
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 PatchEmbed — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a PatchEmbed 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
PatchEmbed Ran
ran on a synthesized input
jaehyunnn/ViTPose_pytorch/models/model.py
code served (permissive licence) · get_code("c26b0690c76a9818")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, class construct-then-call prompt, model not recorded) 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 torch
import torch.nn as nn
from itertools import repeat
import collections.abc

def _ntuple(n):
    def parse(x):
        if isinstance(x, collections.abc.Iterable) and (not isinstance(x, str)):
            return x
        return tuple(repeat(x, n))
    return parse

to_2tuple = _ntuple(2)

class PatchEmbed(nn.Module):
    """ Image to Patch Embedding
    """

    def __init__(self, img_size=224, patch_size=16, in_chans=3, embed_dim=768, ratio=1):
        super().__init__()
        img_size = to_2tuple(img_size)
        patch_size = to_2tuple(patch_size)
        num_patches = img_size[1] // patch_size[1] * (img_size[0] // patch_size[0]) * ratio ** 2
        self.patch_shape = (int(img_size[0] // patch_size[0] * ratio), int(img_size[1] // patch_size[1] * ratio))
        self.origin_patch_shape = (int(img_size[0] // patch_size[0]), int(img_size[1] // patch_size[1]))
        self.img_size = img_size
        self.patch_size = patch_size
        self.num_patches = num_patches
        self.proj = nn.Conv2d(in_chans, embed_dim, kernel_size=patch_size, stride=patch_size[0] // ratio, padding=4 + 2 * (ratio // 2 - 1))

    def forward(self, x, **kwargs):
        B, C, H, W = x.shape
        x = self.proj(x)
        Hp, Wp = (x.shape[2], x.shape[3])
        x = x.flatten(2).transpose(1, 2)
        return (x, (Hp, Wp))


INIT_ARGS = {
    'img_size': 64,
    'patch_size': 8,
    'in_chans': 3,
    'embed_dim': 16,
    'ratio': 1
}

ARGS = {
    'x': torch.randn(2, 3, 64, 64)
}

if __name__ == '__main__':
    model = PatchEmbed(**INIT_ARGS)
    output = model(**ARGS)
    print("Output shape:", output[0].shape)
    print("Patch dimensions:", output[1])
Setup block drafted by
a language model, class construct-then-call prompt, model not recorded
Imports
torch — 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: torch and numpy seeded to 0 by the harness immediately before the class was constructed from INIT_ARGS (fixture_check.py)
_calc_distances Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
JunkyByte/easy_ViTPose/easy_ViTPose/vit_models/model.py
code served (permissive licence) · get_code("c4c1705ff75a3ef5")
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 (3, 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

# Construct valid arguments for _calc_distances

# Batch size: N=2
# Num keypoints: K=3
# Dimension: D=2

N, K, D = 2, 3, 2

# preds: predicted keypoint locations [N, K, D]
preds = np.array([
    [[10.5, 20.3], [15.2, 25.1], [12.0, 18.5]],
    [[11.0, 21.5], [16.0, 26.0], [13.0, 19.5]]
], dtype=np.float32)

# targets: ground truth keypoint locations [N, K, D]
targets = np.array([
    [[10.0, 20.0], [15.0, 25.0], [12.5, 18.0]],
    [[11.5, 21.0], [16.5, 26.5], [12.5, 20.0]]
], dtype=np.float32)

# mask: visibility mask [N, K] - True for visible, False for invisible
mask = np.array([
    [True, True, False],
    [True, False, True]
], dtype=bool)

# normalize: normalization values (typically heatmap size) [N, D]
normalize = np.array([
    [64.0, 64.0],
    [64.0, 64.0]
], dtype=np.float32)

ARGS = {
    "preds": preds,
    "targets": targets,
    "mask": mask,
    "normalize": normalize
}
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 (3, 2) — shape (3, 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
_distance_acc Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
JunkyByte/easy_ViTPose/easy_ViTPose/vit_models/model.py
code served (permissive licence) · get_code("c053b8c18815c5f8")
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 float64, ndim 0, 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 torch

torch.manual_seed(0)

# Create a sample distances array
# The function expects np.ndarray with shape [N,] containing normalized distances
# Some values should be valid (>= 0), some can be -1 (missing/invalid)
distances = np.array([0.2, 0.6, -1, 0.3, 0.8, -1, 0.1, 0.4], dtype=np.float32)

# Threshold parameter
thr = 0.5

ARGS = {
    "distances": distances,
    "thr": thr,
}
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 float64, ndim 0, dtype family float — no output shape is stated: observe_signature.py recorded float ndim 0, which disagrees with the fixture run
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
_gaussian_blur Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
JunkyByte/easy_ViTPose/easy_ViTPose/vit_models/model.py
code served (permissive licence) · get_code("92e7b0a551d548fb")
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 (2, 3, 8, 6) 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 cv2

# Create a small batch of heatmaps with shape (N, K, H, W)
# N=2 (batch_size), K=3 (num_keypoints), H=8 (height), W=6 (width)
batch_size = 2
num_keypoints = 3
height = 8
width = 6

# Generate random heatmaps with values between 0 and 1
heatmaps = np.random.rand(batch_size, num_keypoints, height, width).astype(np.float32)

ARGS = {
    'heatmaps': heatmaps,
    'kernel': 11,
}
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 4, dtype family float, shape (2, 3, 8, 6) — shape (2, 3, 8, 6) 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_max_preds Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
JunkyByte/easy_ViTPose/easy_ViTPose/vit_models/model.py
code served (permissive licence) · get_code("e53d577562ffb351")
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

# Create small heatmaps with shape (N, K, H, W)
# N=2 (batch size), K=3 (num keypoints), H=4 (height), W=5 (width)
heatmaps = np.random.rand(2, 3, 4, 5).astype(np.float32)

ARGS = {
    'heatmaps': heatmaps,
}
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
_taylor Ran
ran; our contract draft was wrong, not the code
JunkyByte/easy_ViTPose/easy_ViTPose/vit_models/model.py
code served (permissive licence) · get_code("6c308dbbb4a9949b")
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 (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 heatmap (H=10, W=12)
heatmap = np.random.rand(10, 12).astype(np.float32)

# Create coordinates in the valid range where the taylor expansion can be applied
# The condition checks: 1 < px < W-2 and 1 < py < H-2
# So px must be in range [2, 9] and py must be in range [2, 7]
coord = np.array([5.0, 4.0], dtype=np.float32)

ARGS = {
    "heatmap": heatmap,
    "coord": coord,
}
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 (2,) — shape (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
keypoint_pck_accuracy Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
JunkyByte/easy_ViTPose/easy_ViTPose/vit_models/model.py
code served (permissive licence) · get_code("8d099da3b789e317")
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

np.random.seed(0)

# batch_size: N = 2
# num_keypoints: K = 4
# dimension: D = 2

N, K, D = 2, 4, 2

# Predicted keypoint locations [N, K, D]
pred = np.random.rand(N, K, D).astype(np.float32) * 100

# Ground truth keypoint locations [N, K, D]
gt = np.random.rand(N, K, D).astype(np.float32) * 100

# Visibility mask [N, K]
mask = np.ones((N, K), dtype=bool)
mask[0, 1] = False  # Make one keypoint invisible

# Threshold for PCK
thr = 0.5

# Normalization factor [N, D] - typical value is heatmap_size
normalize = np.array([[64.0, 48.0], [64.0, 48.0]], dtype=np.float32)

ARGS = {
    'pred': pred,
    'gt': gt,
    'mask': mask,
    'thr': thr,
    'normalize': normalize
}
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
post_dark_udp Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
JunkyByte/easy_ViTPose/easy_ViTPose/vit_models/model.py
code served (permissive licence) · get_code("8294c29e3ad40443")
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 (2, 3, 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 torch

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

# Define small test data
# N: num persons, K: num keypoints, H: height, W: width, B: batch size
N = 2  # num persons
K = 3  # num keypoints
H = 4  # height of heatmaps
W = 5  # width of heatmaps
B = 1  # batch size (can be 1 or N)

# Create initial coordinates: [N, K, 2]
# Coordinates should be within the heatmap bounds
coords = np.random.uniform(0.5, W - 1.5, size=(N, K, 2)).astype(np.float32)
coords[..., 0] = np.clip(coords[..., 0], 0, W - 1)
coords[..., 1] = np.clip(coords[..., 1], 0, H - 1)

# Create batch heatmaps: [B, K, H, W]
# Heatmaps should be positive values
batch_heatmaps = np.random.uniform(1.0, 10.0, size=(B, K, H, W)).astype(np.float32)

# Kernel size for Gaussian blur (must be odd)
kernel = 3

ARGS = {
    "coords": coords,
    "batch_heatmaps": batch_heatmaps,
    "kernel": kernel,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
cv2, 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 3, dtype family float, shape (2, 3, 2) — shape (2, 3, 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
trunc_normal_ Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
JunkyByte/easy_ViTPose/easy_ViTPose/vit_models/model.py
code served (permissive licence) · get_code("c3487124f762d881")
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, 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 math

torch.manual_seed(0)

# Create a small tensor to initialize with truncated normal
tensor = torch.empty(2, 4)

# Define arguments for trunc_normal_
ARGS = {
    'tensor': tensor,
    'mean': 0.0,
    'std': 1.0,
    'a': -2.0,
    'b': 2.0
}
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, 4) — shape (2, 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
TopdownHeatmapBaseHead Not yet run
unverified (harvested, no recorded run)
JunkyByte/easy_ViTPose/easy_ViTPose/vit_models/model.py
code served (permissive licence) · get_code("c828e7e8249b42be")
How it ran

no invocation recorded.

TopdownHeatmapSimpleHead Not yet run
unverified (harvested, no recorded run)
JunkyByte/easy_ViTPose/easy_ViTPose/vit_models/model.py
code served (permissive licence) · get_code("bbca4a34f6f76844")
How it ran

no invocation recorded.

TopdownHeatmapSimpleHead Not yet run
unverified (harvested, no recorded run)
jaehyunnn/ViTPose_pytorch/models/model.py
code served (permissive licence) · get_code("dbcb1917083518a7")
How it ran

no invocation recorded.

ViT Not yet run
unverified (harvested, no recorded run)
gpastal24/ViTPose-Pytorch/src/vitpose_infer/builder/backbones/vit.py
pointer only (licence: GPL-3.0) · get_code("d17daecae39897ce")
How it ran

The body of this sample is withheld: GPL-3.0, as GitHub detects it on this paper's repository, gpastal24/ViTPose-Pytorch, is not in the permissive set Syntology serves inline. 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.

ViT Not yet run
unverified (harvested, no recorded run)
JunkyByte/easy_ViTPose/easy_ViTPose/vit_models/model.py
code served (permissive licence) · get_code("ec2566dcbcbac9aa")
How it ran

no invocation recorded.

ViT Not yet run
unverified (harvested, no recorded run)
jaehyunnn/ViTPose_pytorch/models/model.py
code served (permissive licence) · get_code("a3947ec0fada50f7")
How it ran

no invocation recorded.

ViTPose Not yet run
unverified (harvested, no recorded run)
JunkyByte/easy_ViTPose/easy_ViTPose/vit_models/model.py
code served (permissive licence) · get_code("e38694c775a09153")
How it ran

no invocation recorded.

ViTPose Not yet run
unverified (harvested, no recorded run)
jaehyunnn/ViTPose_pytorch/models/model.py
code served (permissive licence) · get_code("ee523b19fa42ef3e")
How it ran

no invocation recorded.

keypoints_from_heatmaps Not yet run
unverified (harvested, no recorded run)
JunkyByte/easy_ViTPose/easy_ViTPose/vit_models/model.py
code served (permissive licence) · get_code("c706a215373869f8")
How it ran

no invocation recorded.

normal_init Not yet run
unverified (harvested, no recorded run)
JunkyByte/easy_ViTPose/easy_ViTPose/vit_models/model.py
code served (permissive licence) · get_code("fb206aa7be7aa229")
How it ran

no invocation recorded.

pose_pck_accuracy Not yet run
unverified (harvested, no recorded run)
JunkyByte/easy_ViTPose/easy_ViTPose/vit_models/model.py
code served (permissive licence) · get_code("6d08729cda37113c")
How it ran

no invocation recorded.

resize Not yet run
unverified (harvested, no recorded run)
JunkyByte/easy_ViTPose/easy_ViTPose/vit_models/model.py
code served (permissive licence) · get_code("471e5bd20e25b2a2")
How it ran

no invocation recorded.

resize Not yet run
unverified (harvested, no recorded run)
jaehyunnn/ViTPose_pytorch/models/model.py
code served (permissive licence) · get_code("126e704199014f2d")
How it ran

no invocation recorded.

Also present in other papers' code

Of 31 samples lifted from this paper's linked repositories, 31 carry a function identity and 0 do not. Of the 25 distinct functions with an identity, 6 are also present in other papers whose code we hold (2 with at least one paper dated earlier, 0 the same month, 6 later) and 19 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
trunc_normal_
brick of this paper's assembly: helper
6 other papers
5 later, 1 undated
2308.01045 (2023-08)
Dynamic Token Pruning in Plain Vision Transformers for Semantic Segmentation
2311.02326 (2023-11)
FragXsiteDTI: Revealing Responsible Segments in Drug-Target Interaction with Transformer-Driven Interpretation
2311.07788 (2023-11)
CSLP-AE: A Contrastive Split-Latent Permutation Autoencoder Framework for Zero-Shot Electroencephalography Signal Conversion
and 3 more not named here
pose_pck_accuracy
brick of this paper's assembly: helper
3 other papers
3 later
2312.10758 (2023-12)
SHaRPose: Sparse High-Resolution Representation for Human Pose Estimation
2403.01813 (2024-03)
A Simple Baseline for Efficient Hand Mesh Reconstruction
2411.16289 (2024-11)
Utilizing Uncertainty in 2D Pose Detectors for Probabilistic 3D Human Mesh Recovery
DropPath
brick of this paper's assembly: helper
2 other papers
1 dated earlier, 1 later · 1 only as an assembly brick
2012.12877 (2020-12)
Training data-efficient image transformers & distillation through attention
2303.05675 (2023-03)
HumanBench: Towards General Human-centric Perception with Projector Assisted Pretraining
HybridEmbed
brick of this paper's assembly: helper
2 other papers
1 dated earlier, 1 later · 1 only as an assembly brick
2010.11929 (2020-10)
An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale
2207.10447 (2022-07)
Weakly Supervised Object Localization via Transformer with Implicit Spatial Calibration
keypoint_pck_accuracy
brick of this paper's assembly: helper
2 other papers
2 later
2403.01813 (2024-03)
A Simple Baseline for Efficient Hand Mesh Reconstruction
2411.16289 (2024-11)
Utilizing Uncertainty in 2D Pose Detectors for Probabilistic 3D Human Mesh Recovery
Attention
brick of this paper's assembly: helper
1 other paper
1 later · 1 only as an assembly brick
2303.14189 (2023-03)
FastViT: A Fast Hybrid Vision Transformer using Structural Reparameterization

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

Although no specific domain knowledge is considered in the design, plain vision transformers have shown excellent performance in visual recognition tasks. However, little effort has been made to reveal the potential of such simple structures for pose estimation tasks. In this paper, we show the surprisingly good capabilities of plain vision transformers for pose estimation from various aspects, namely simplicity in model structure, scalability in model size, flexibility in training paradigm, and transferability of knowledge between models, through a simple baseline model called ViTPose. Specifically, ViTPose employs plain and non-hierarchical vision transformers as backbones to extract features for a given person instance and a lightweight decoder for pose estimation. It can be scaled up from 100M to 1B parameters by taking the advantages of the scalable model capacity and high parallelism of transformers, setting a new Pareto front between throughput and performance. Besides, ViTPose is very flexible regarding the attention type, input resolution, pre-training and finetuning strategy, as well as dealing with multiple pose tasks. We also empirically demonstrate that the knowledge of large ViTPose models can be easily transferred to small ones via a simple knowledge token. Experimental results show that our basic ViTPose model outperforms representative methods on the challenging MS COCO Keypoint Detection benchmark, while the largest model sets a new state-of-the-art, i.e., 80.9 AP on the MS COCO test-dev set. The code and models are available at https://github.com/ViTAE-Transformer/ViTPose.

For agents

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

get_harvested_code_for_paper("2204.12484")
get_code_for_paper("2204.12484")
have("2204.12484")

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

curl https://syntology.ai/api/ran/2204.12484.json

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

[![Syntology run record](https://syntology.ai/api/ran/2204.12484.svg)](https://syntology.ai/paper/2204.12484)

Connect an agent — have() is free.