SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · aaai_28166 · AAAI · 2024

Keypoint Fusion for RGB-D Based 3D Hand Pose Estimation

Qi Qi, Jingyu Wang, Haifeng Sun, Xingyu Liu, Zirui Zhuang, Jianxin Liao, Yuanyuan Gao, Pengfei Ren

Open in the Atlas

Code that ran

14 of 16 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 6 have no contract verdict; 8 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/aaai_28166.json · badge: /api/ran/aaai_28166.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
ru1ven/KeypointFusion canonical (classifier label)
classifier basis: census
14 ran · 2 no recorded run
FunctionStatusWhere it lives
batch_rodrigues Ran
ran; our contract draft was wrong, not the code
ru1ven/KeypointFusion/model/mano_head.py
pointer only (licence: MIT); the stored copy get_code serves, from heartstrive/handgcat, is pointer only (licence: NONE) · get_code("849f989bd754317e")
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 heartstrive/handgcat, 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.

conv1x1 Ran
ran; our contract draft was wrong, not the code
ru1ven/KeypointFusion/model/centerNet.py
code served (permissive licence) · get_code("d9def42110729a85")
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 Conv2d. 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)

# The function conv1x1 takes in_planes, out_planes, and stride as parameters
# It returns an nn.Conv2d layer configured for 1x1 convolution
# We need to provide valid integer arguments for these parameters

ARGS = {
    "in_planes": 3,
    "out_planes": 64,
    "stride": 1
}
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 Conv2d — a Conv2d output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a Conv2d 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
conv1x1 Ran
ran; our contract draft was wrong, not the code
ru1ven/KeypointFusion/model/resnet.py
pointer only (licence: MIT); the stored copy get_code serves, from medaitech/scrd4ad, is pointer only (licence: NONE) · get_code("2a80220dabcb742a")
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 medaitech/scrd4ad, 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.

conv3x3 Ran
ran; our contract draft was wrong, not the code
ru1ven/KeypointFusion/model/centerNet.py
code served (permissive licence) · get_code("160bb14bd76201b4")
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 Conv2d. 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 = {
    "in_planes": 3,
    "out_planes": 64,
    "stride": 1,
    "groups": 1,
    "dilation": 1,
}
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 Conv2d — a Conv2d output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a Conv2d 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
conv3x3 Ran
ran; our contract draft was wrong, not the code
ru1ven/KeypointFusion/model/resnet.py
pointer only (licence: MIT); the stored copy get_code serves, from medaitech/scrd4ad, is pointer only (licence: NONE) · get_code("600ff2c45e0de056")
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 medaitech/scrd4ad, 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.

conv_bn_relu Ran
ran on a synthesized input
ru1ven/KeypointFusion/convNeXT/resnetUnet.py
code served (permissive licence) · get_code("ed055d3f8aa61273")
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 Sequential. 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 = {
    'in_planes': 3,
    'out_planes': 64,
    'kernel': 3
}
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 Sequential — a Sequential output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a Sequential output the recorded ndim 1 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
convtranspose_bn_relu Ran
ran on a synthesized input
ru1ven/KeypointFusion/convNeXT/resnetUnet.py
code served (permissive licence) · get_code("a7c667c4fe44f3d1")
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 Sequential. 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 = {
    "in_planes": 3,
    "out_planes": 64,
    "kernel": 4
}
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 Sequential — a Sequential output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a Sequential output the recorded ndim 1 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
joint2offset Ran
ran on a synthesized input
ru1ven/KeypointFusion/model/model.py
code served (permissive licence) · get_code("51905d976cfde989")
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 4, dtype family float, shape (2, 12, 8, 8) 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.functional as F

torch.manual_seed(0)

# Construct arguments for joint2offset function
batch_size = 2
feature_size = 8
img_height = 16
img_width = 16
joint_num = 3
kernel_size = 5.0

# joint: tensor of shape (batch_size, joint_num, 3) representing joint coordinates
# The function expects it to be reshapeable to (batch_size, joint_num, 3)
joint = torch.randn(batch_size, joint_num, 3)

# img: tensor of shape (batch_size, channels, img_height, img_width)
# The function interpolates it and concatenates to coords
# It also checks img.lt(0.99) for masking, so values should be in valid range
img = torch.randn(batch_size, 1, img_height, img_width) * 0.5

ARGS = {
    "joint": joint,
    "img": img,
    "kernel_size": kernel_size,
    "feature_size": feature_size,
}
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 4, dtype family float, shape (2, 12, 8, 8) — shape (2, 12, 8, 8) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
load_dualpath_model Ran
ran on a synthesized input
ru1ven/KeypointFusion/model/resnet.py
code served (permissive licence) · get_code("87588727fef4a4ab")
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 DualPathModel. 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 collections import OrderedDict

torch.manual_seed(0)

# Build a minimal nn.Module that has the structure expected by load_dualpath_model
class DualPathModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.conv1 = nn.Conv2d(3, 4, kernel_size=3)
        self.bn1 = nn.BatchNorm2d(4)
        self.conv2 = nn.Conv2d(4, 8, kernel_size=3)
        self.bn2 = nn.BatchNorm2d(8)
        self.conv3 = nn.Conv2d(8, 16, kernel_size=3)
        self.bn3 = nn.BatchNorm2d(16)
        self.downsample = nn.Sequential(
            nn.Conv2d(3, 4, kernel_size=1),
            nn.BatchNorm2d(4)
        )
        self.depth_conv1 = nn.Conv2d(3, 4, kernel_size=3)
        self.depth_bn1 = nn.BatchNorm2d(4)
        self.depth_conv2 = nn.Conv2d(4, 8, kernel_size=3)
        self.depth_bn2 = nn.BatchNorm2d(8)
        self.depth_conv3 = nn.Conv2d(8, 16, kernel_size=3)
        self.depth_bn3 = nn.BatchNorm2d(16)
        self.depth_downsample = nn.Sequential(
            nn.Conv2d(3, 4, kernel_size=1),
            nn.BatchNorm2d(4)
        )
    
    def forward(self, x):
        return x

model = DualPathModel()

# Create a state dict that mimics what would be loaded from a checkpoint
raw_state_dict = OrderedDict()
raw_state_dict['conv1.weight'] = torch.randn(4, 3, 3, 3)
raw_state_dict['conv1.bias'] = torch.randn(4)
raw_state_dict['bn.1.weight'] = torch.randn(4)
raw_state_dict['bn.1.bias'] = torch.randn(4)
raw_state_dict['conv2.weight'] = torch.randn(8, 4, 3, 3)
raw_state_dict['conv2.bias'] = torch.randn(8)
raw_state_dict['bn.2.weight'] = torch.randn(8)
raw_state_dict['bn.2.bias'] = torch.randn(8)
raw_state_dict['conv3.weight'] = torch.randn(16, 8, 3, 3)
raw_state_dict['conv3.bias'] = torch.randn(16)
raw_state_dict['bn.3.weight'] = torch.randn(16)
raw_state_dict['bn.3.bias'] = torch.randn(16)
raw_state_dict['downsample.0.weight'] = torch.randn(4, 3, 1, 1)
raw_state_dict['downsample.0.bias'] = torch.randn(4)

ARGS = {
    'model': model,
    'model_file': raw_state_dict,
    'is_restore': False
}
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 DualPathModel — a DualPathModel output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a DualPathModel 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
logsumexp_2d Ran
ran; our contract draft was wrong, not the code
ru1ven/KeypointFusion/model/cbam.py
code served (permissive licence) · get_code("7abfa32a2f1d7424")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Sonnet 4.5, prompt python_fixture_v1) and returned Tensor, ndim 3, dtype family float, shape (2, 3, 1) 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 3D or higher dimensional tensor
# It accesses tensor.size(0), tensor.size(1), and flattens the rest
# Let's create a 3D tensor with shape (batch=2, dim1=3, dim2=4)
tensor = torch.randn(2, 3, 4)

ARGS = {
    "tensor": tensor
}
Setup block drafted by
a language model, Claude Sonnet 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 Tensor, ndim 3, dtype family float, shape (2, 3, 1) — shape (2, 3, 1) 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
offset2joint_softmax Ran
ran on a synthesized input
ru1ven/KeypointFusion/model/hourglass.py
code served (permissive licence) · get_code("09e012a9703e863b")
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 (2, 4, 3) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn.functional as F

torch.manual_seed(0)

# Analyze the function to understand required tensor shapes and types:
# - offset: shape (batch_size, joint_num * 4, feature_size, feature_size)
#   The function divides joint_num by 4, so joint_num must be divisible by 4
#   offset is split into offset_unit (joint_num * 3 channels) and heatmap (joint_num channels)
# - depth: shape (batch_size, 1, height, width) or will be interpolated to (batch_size, 1, feature_size, feature_size)
# - kernel_size: scalar, used in arithmetic operations
# - scale: scalar with default 30, used to scale heatmap

batch_size = 2
feature_size = 4
joint_num = 4  # Must be divisible by 4, so 4 joints means 16 channels total
total_channels = joint_num * 4  # 16 channels: 12 for offset_unit, 4 for heatmap

# offset tensor: (batch_size, joint_num*4, feature_size, feature_size)
offset = torch.randn(batch_size, total_channels, feature_size, feature_size)

# depth tensor: (batch_size, 1, feature_size, feature_size)
# Must have values < 0.99 to create non-zero mask
depth = torch.rand(batch_size, 1, feature_size, feature_size) * 0.9

# kernel_size: scalar value
kernel_size = 5

# scale: default is 30, but can be provided
scale = 30

ARGS = {
    "offset": offset,
    "depth": depth,
    "kernel_size": kernel_size,
    "scale": scale,
}
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 (2, 4, 3) — shape (2, 4, 3) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
offset2joint_weight Ran
ran on a synthesized input
ru1ven/KeypointFusion/model/model.py
code served (permissive licence) · get_code("92a607bd02e3111b")
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 (2, 3, 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 torch
import torch.nn.functional as F

torch.manual_seed(0)

# Construct valid arguments for offset2joint_weight
batch_size = 2
feature_size = 4
joint_num = 3

# offset: shape [batch_size, joint_num * 5, feature_size, feature_size]
# joint_num * 5 = 15 (3 for offset_unit, 1 for heatmap, 1 for weight)
offset = torch.randn(batch_size, joint_num * 5, feature_size, feature_size)

# depth: shape [batch_size, 1, feature_size, feature_size]
depth = torch.rand(batch_size, 1, feature_size, feature_size)

# kernel_size: can be either a scalar or a tensor
# Based on the function, it's used in arithmetic operations
# Let's use a tensor of shape [joint_num]
kernel_size = torch.tensor([1.5, 2.0, 2.5])

ARGS = {
    "offset": offset,
    "depth": depth,
    "kernel_size": kernel_size
}
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 (2, 3, 3) — shape (2, 3, 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
quat2aa Ran
ran; our contract draft was wrong, not the code
ru1ven/KeypointFusion/model/mano_head.py
pointer only (licence: MIT); the stored copy get_code serves, from amathislab/HOISDF, is pointer only (licence: NONE) · get_code("40dc9f5943418548")
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 amathislab/HOISDF, 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.

quat2mat Ran
ran; our contract draft was wrong, not the code
ru1ven/KeypointFusion/model/mano_head.py
pointer only (licence: MIT); the stored copy get_code serves, from heartstrive/handgcat, is pointer only (licence: NONE) · get_code("be0f652ebe3d9098")
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 heartstrive/handgcat, 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.

img2pcl Not yet run
unverified (harvested, no recorded run)
ru1ven/KeypointFusion/model/model.py
code served (permissive licence) · get_code("7a1f0fd87501fe34")
How it ran

no invocation recorded.

joint2offset Not yet run
unverified (harvested, no recorded run)
ru1ven/KeypointFusion/model/hourglass.py
code served (permissive licence) · get_code("9d934f04fe74e54e")
How it ran

no invocation recorded.

Also present in other papers' code

Of 16 samples lifted from this paper's linked repositories, 16 carry a function identity and 0 do not. Of the 16 distinct functions with an identity, 8 are also present in other papers whose code we hold (7 with at least one paper dated earlier, 0 the same month, 5 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
conv1x1 431 other papers
313 dated earlier, 30 later, 87 same year, order not determinable, 1 undated · 38 only as an assembly brick
1409.1556 (2014-09)
Very Deep Convolutional Networks for Large-Scale Image Recognition
1506.01186 (2015-06)
Cyclical Learning Rates for Training Neural Networks
1507.05717 (2015-07)
An End-to-End Trainable Neural Network for Image-based Sequence Recognition and Its Application to Scene Text Recognition
and 428 more not named here
conv3x3 312 other papers
214 dated earlier, 22 later, 75 same year, order not determinable, 1 undated · 31 only as an assembly brick
1409.1556 (2014-09)
Very Deep Convolutional Networks for Large-Scale Image Recognition
1507.05717 (2015-07)
An End-to-End Trainable Neural Network for Image-based Sequence Recognition and Its Application to Scene Text Recognition
1512.03385 (2015-12)
Deep Residual Learning for Image Recognition
and 309 more not named here
conv1x1 112 other papers
64 dated earlier, 10 later, 38 same year, order not determinable · 6 only as an assembly brick
1512.03385 (2015-12)
Deep Residual Learning for Image Recognition
1806.01531 (2018-06)
Deep Mixture of Experts via Shallow Embedding
1905.12794 (2019-05)
Fashion IQ: A New Dataset Towards Retrieving Images by Natural Language Feedback
and 109 more not named here
conv3x3 111 other papers
65 dated earlier, 9 later, 37 same year, order not determinable · 6 only as an assembly brick
1512.03385 (2015-12)
Deep Residual Learning for Image Recognition
1806.01531 (2018-06)
Deep Mixture of Experts via Shallow Embedding
1905.12794 (2019-05)
Fashion IQ: A New Dataset Towards Retrieving Images by Natural Language Feedback
and 108 more not named here
logsumexp_2d 13 other papers
10 dated earlier, 1 later, 2 same year, order not determinable
1806.07366 (2018-06)
Neural Ordinary Differential Equations
1807.06521 (2018-07)
CBAM: Convolutional Block Attention Module
1810.12348 (2018-10)
Gather-Excite: Exploiting Feature Context in Convolutional Neural Networks
and 10 more not named here
quat2mat 7 other papers
5 dated earlier, 2 same year, order not determinable
1905.03244 (2019-05)
Convolutional Mesh Regression for Single-Image Human Shape Reconstruction
2004.01946 (2020-04)
Weakly-Supervised Mesh-Convolutional Hand Reconstruction in the Wild
2006.05734 (2020-06)
3D Human Mesh Regression with Dense Correspondence
and 4 more not named here
batch_rodrigues 3 other papers
1 dated earlier, 2 same year, order not determinable
2108.06819 (2021-08)
EventHPE: Event-based 3D Human Pose and Shape Estimation
2402.17062 (2024-02)
HOISDF: Constraining 3D Hand-Object Pose Estimation with Global Signed Distance Fields
2403.07912 (2024-03)
HandGCAT: Occlusion-Robust 3D Hand Mesh Reconstruction from Monocular Images
quat2aa 2 other papers
2 same year, order not determinable
2402.17062 (2024-02)
HOISDF: Constraining 3D Hand-Object Pose Estimation with Global Signed Distance Fields
2403.07912 (2024-03)
HandGCAT: Occlusion-Robust 3D Hand Mesh Reconstruction from Monocular Images

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

Previous 3D hand pose estimation methods primarily rely on a single modality, either RGB or depth, and the comprehensive utilization of the dual modalities has not been extensively explored. RGB and depth data provide complementary information and thus can be fused to enhance the robustness of 3D hand pose estimation. However, there exist two problems for applying existing fusion methods in 3D hand pose estimation: redundancy of dense feature fusion and ambiguity of visual features. First, pixel-wise feature interactions introduce high computational costs and ineffective calculations of invalid pixels. Second, visual features suffer from ambiguity due to color and texture similarities, as well as depth holes and noise caused by frequent hand movements, which interferes with modeling cross-modal correlations. In this paper, we propose Keypoint-Fusion for RGB-D based 3D hand pose estimation, which leverages the unique advantages of dual modalities to mutually eliminate the feature ambiguity, and performs cross-modal feature fusion in a more efficient way. Specifically, we focus cross-modal fusion on sparse yet informative spatial regions (i.e. keypoints). Meanwhile, by explicitly extracting relatively more reliable information as disambiguation evidence, depth modality provides 3D geometric information for RGB feature pixels, and RGB modality complements the precise edge information lost due to the depth noise. Keypoint-Fusion achieves state-of-the-art performance on two challenging hand datasets, significantly decreasing the error compared with previous single-modal methods.

For agents

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

get_harvested_code_for_paper("aaai_28166")
get_code_for_paper("aaai_28166")
have("aaai_28166")

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

curl https://syntology.ai/api/ran/aaai_28166.json

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

[![Syntology run record](https://syntology.ai/api/ran/aaai_28166.svg)](https://syntology.ai/paper/aaai_28166)

Connect an agent — have() is free.