SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2303.08658 · CVPR · 2023

Skinned Motion Retargeting with Residual Perception of Motion Semantics & Geometry

Jue Wang, Ying Shan, Shaoli Huang, Fang Zhao, Di Kang, Junwu Weng, Zhigang Tu, Linchao Bao, Xuefei Zhe, Jiaxu Zhang

arXiv · PDF · Open in the Atlas

Code that ran

11 of 14 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 11 have no contract verdict; 0 failures of our instrument; 3 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/2303.08658.json · badge: /api/ran/2303.08658.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
Kebii/R2ET canonical (classifier label)
classifier basis not recorded on the link
11 ran · 3 no recorded run
FunctionStatusWhere it lives
get_bounding_boxes Ran
ran on a synthesized input
Kebii/R2ET/src/model_pmnet.py
code served (permissive licence) · get_code("941fcb7770df6b30")
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, 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 torch

torch.manual_seed(0)

# The function expects vertices to be a tensor with shape (num_people, num_vertices, 3)
# where each vertex has 3 coordinates (x, y, z)
# It computes min and max along dimension 1 (the vertices dimension)

# Create a small tensor with:
# - 2 people (num_people = 2)
# - 4 vertices per person (num_vertices = 4)
# - 3 coordinates per vertex (x, y, z)
vertices = torch.randn(2, 4, 3)

ARGS = {
    "vertices": vertices
}
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, 2, 3) — shape (2, 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_floor Ran
ran on a synthesized input
Kebii/R2ET/src/utils.py
code served (permissive licence) · get_code("7dcd2930edbaffd4")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned float32, ndim 0, dtype family float, shape () as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np

# Analyze the function:
# - get_floor takes sk_in as parameter
# - sk_in is indexed as sk_in[0, i, 1] where i ranges from 0 to 13
# - So sk_in needs shape at least (1, 14, 2)
# - The function accesses indices: 0, 6, 7, 8, 9, 10, 11, 12, 13 on dimension 1
# - And always index 1 on dimension 2
# - We need dimension 2 to have at least 2 elements (indices 0 and 1)

# Create a small valid sk_in array
# Shape: (batch=1, joints=14, coordinates=2)
sk_in = np.random.randn(1, 14, 2).astype(np.float32)

# Ensure we have valid numeric values
sk_in[0, 0, 1] = 1.0  # h_root
sk_in[0, 6, 1] = 0.1
sk_in[0, 7, 1] = 0.1
sk_in[0, 8, 1] = 0.1
sk_in[0, 9, 1] = 0.1
sk_in[0, 10, 1] = 0.15
sk_in[0, 11, 1] = 0.15
sk_in[0, 12, 1] = 0.15
sk_in[0, 13, 1] = 0.15

ARGS = {
    "sk_in": sk_in
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type float32, ndim 0, dtype family float, shape () — shape () as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
get_height Ran
ran on a synthesized input
Kebii/R2ET/inference_pmnet.py
code served (permissive licence) · get_code("66c1c18735de54f7")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned float32, ndim 0, dtype family float, shape () as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np

# The function expects joints to be a numpy array with shape (10, 3) or similar
# It accesses joints[5], joints[4], etc. (10 different indices: 0-9)
# Each joint has coordinates (x, y, z) - accessing with [:] to get all coordinates
# The function computes distances between consecutive joints in a skeleton chain

joints = np.array([
    [0.0, 0.0, 0.0],      # joint 0
    [0.1, 0.1, 0.1],      # joint 1
    [0.2, 0.2, 0.2],      # joint 2
    [0.3, 0.3, 0.3],      # joint 3
    [0.4, 0.4, 0.4],      # joint 4
    [0.5, 0.5, 0.5],      # joint 5
    [0.6, 0.6, 0.6],      # joint 6
    [0.7, 0.7, 0.7],      # joint 7
    [0.8, 0.8, 0.8],      # joint 8
    [0.9, 0.9, 0.9],      # joint 9
], dtype=np.float32)

ARGS = {
    'joints': joints
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type float32, ndim 0, dtype family float, shape () — shape () as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
get_skel Ran
ran on a synthesized input
Kebii/R2ET/inference_bvh.py
code served (permissive licence) · get_code("da9518aab5e9f154")
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 (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 numpy as np

# Create sample data for joints and parents
# joints: shape (num_joints, num_features)
# parents: shape (num_joints,) where each element is the parent joint index or -1 for root

num_joints = 4
num_features = 3

joints = np.random.randn(num_joints, num_features).astype(np.float32)
parents = np.array([-1, 0, 1, 2], dtype=np.int32)

ARGS = {
    "joints": joints,
    "parents": parents,
}
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 (4, 3) — shape (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
linear_blend_skinning Ran
ran on a synthesized input
Kebii/R2ET/src/linear_blend_skin.py
code served (permissive licence) · get_code("8ba835eb9aa421e9")
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, 8, 3) as a later run recorded it. 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, 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:
# - parents: list of parent indices for bones
# - quat: (T, 22, 4) quaternions
# - rest_skel: (22, 3) rest skeleton positions
# - mesh_vertices: (vertex_num, 3) vertex positions
# - skin_weights: (vertex_num, 22) skinning weights

T = 2  # time steps
num_bones = 22
num_vertices = 8

# parents: list mapping each bone to its parent (0-indexed, parent of root is 0)
parents = [0, 0, 1, 2, 3, 1, 5, 6, 1, 8, 9, 10, 8, 12, 13, 8, 15, 16, 0, 18, 19, 20]

# quat: (T=2, 22, 4) - quaternions normalized
quat = torch.randn(T, num_bones, 4)
quat = quat / torch.norm(quat, dim=-1, keepdim=True)

# rest_skel: (22, 3) - rest skeleton positions
rest_skel = torch.randn(num_bones, 3)

# mesh_vertices: (num_vertices=8, 3) - vertex positions in rest pose
mesh_vertices = torch.randn(num_vertices, 3)

# skin_weights: (num_vertices=8, 22) - normalized weights per vertex
skin_weights = torch.rand(num_vertices, num_bones)
skin_weights = skin_weights / skin_weights.sum(dim=1, keepdim=True)

ARGS = {
    "parents": parents,
    "quat": quat,
    "rest_skel": rest_skel,
    "mesh_vertices": mesh_vertices,
    "skin_weights": skin_weights,
}
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, 8, 3) — shape (2, 8, 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
lrelu Ran
ran on a synthesized input
Kebii/R2ET/src/ops.py
code served (permissive licence) · get_code("9acfdca54949c838")
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

torch.manual_seed(0)

# Create a small tensor argument
x = torch.randn(2, 4)

# Create ARGS dict with all parameters
ARGS = {
    "x": x,
    "leak": 0.2
}
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
normalized Ran
ran on a synthesized input
Kebii/R2ET/src/model_pmnet.py
code served (permissive licence) · get_code("ecc81a376cd56687")
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

torch.manual_seed(0)

# The function takes angles and normalizes them by their L2 norm
# angles should be a tensor with shape (..., num_features)
# The function computes lengths along the last dimension and divides

angles = torch.randn(2, 4)  # batch_size=2, feature_dim=4

ARGS = {
    "angles": angles
}
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
q_mul_v Ran
ran on a synthesized input
Kebii/R2ET/src/ops.py
code served (permissive licence) · get_code("714ab5d9cc5a66cc")
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, 1, 3) as a later run recorded it. 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, 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)

def q_mul_q(a, b):
    sqs, oqs = q_broadcast(a, b)
    q0 = sqs[..., 0:1]
    q1 = sqs[..., 1:2]
    q2 = sqs[..., 2:3]
    q3 = sqs[..., 3:4]
    r0 = oqs[..., 0:1]
    r1 = oqs[..., 1:2]
    r2 = oqs[..., 2:3]
    r3 = oqs[..., 3:4]
    qs0 = r0 * q0 - r1 * q1 - r2 * q2 - r3 * q3
    qs1 = r0 * q1 + r1 * q0 - r2 * q3 + r3 * q2
    qs2 = r0 * q2 + r1 * q3 + r2 * q0 - r3 * q1
    qs3 = r0 * q3 - r1 * q2 + r2 * q1 + r3 * q0
    return torch.cat([qs0, qs1, qs2, qs3], dim=-1)

def q_neg(a):
    return a * torch.tensor([[[1, -1, -1, -1]]])

def q_broadcast(sqs, oqs):
    if int(sqs.shape[-2]) == 1:
        sqsn = []
        for l in range(oqs.shape[-2]):
            sqsn.append(sqs)
        sqs = torch.cat(sqsn, dim=-2)
    if int(oqs.shape[-2]) == 1:
        oqsn = []
        for l in range(sqs.shape[-2]):
            oqsn.append(oqs)
        oqs = torch.cat(oqsn, dim=-2)
    return (sqs, oqs)

def q_mul_v(a, b):
    vs = torch.cat([torch.zeros(list(b.shape[:-1]) + [1]), b], dim=-1)
    return q_mul_q(a, q_mul_q(vs, q_neg(a)))[..., 1:4]

a = torch.randn(2, 1, 4)
b = torch.randn(2, 1, 3)

ARGS = {
    "a": a,
    "b": b
}
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, 1, 3) — shape (2, 1, 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
repr6d2mat Ran
ran on a synthesized input
Kebii/R2ET/src/linear_blend_skin.py
code served (permissive licence) · get_code("b8c43b0275e11be8")
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

torch.manual_seed(0)

# The function takes a 6D representation and converts it to a 3x3 rotation matrix
# repr should be a tensor with shape (..., 6) where the last dimension has 6 elements
# The function operates on the last dimension, so we can have any batch shape

# Create a small tensor with shape (2, 6) - batch of 2, 6D representations
repr_tensor = torch.randn(2, 6)

ARGS = {
    "repr": repr_tensor
}
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
softmin Ran
ran on a synthesized input
Kebii/R2ET/inference_bvh.py
code served (permissive licence) · get_code("41b8ced705fecdda")
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

# softmin(x, **kw) calls softmax(-x, **kw)
# softmax(x, **kw) uses:
#   - kw.pop('softness', 1.0) to extract softness parameter
#   - np.max(x, **kw) to find max
#   - np.min(x, **kw) to find min
#   - np.exp() for exponential
#   - np.log() for logarithm
# So x must be a numpy array, and **kw can include keyword arguments for np.max/np.min

# Create a small numpy array
x = np.array([[1.0, 2.0, 3.0, 4.0],
              [5.0, 6.0, 7.0, 8.0]])

# Build ARGS dict with all parameters
ARGS = {
    'x': x,
}
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 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
transforms_rotations Ran
ran on a synthesized input
Kebii/R2ET/src/linear_blend_skin.py
code served (permissive licence) · get_code("fec8a6d5d7231dab")
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

torch.manual_seed(0)

# The function takes rotations as a tensor
# It expects shape (..., 4) where the last dimension contains quaternion components [qw, qx, qy, qz]
# Let's create a small batch of quaternions with shape (2, 4)
rotations = torch.randn(2, 4)

ARGS = {
    "rotations": rotations
}
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
euler_y Not yet run
unverified (harvested, no recorded run)
Kebii/R2ET/src/model_pmnet.py
code served (permissive licence) · get_code("0f51c05d2a21a6fd")
How it ran

no invocation recorded.

get_vels Not yet run
unverified (harvested, no recorded run)
Kebii/R2ET/src/ops.py
code served (permissive licence) · get_code("db36b68f87e646a5")
How it ran

no invocation recorded.

softmax Not yet run
unverified (harvested, no recorded run)
Kebii/R2ET/inference_bvh.py
code served (permissive licence) · get_code("645e56f7684d87a0")
How it ran

no invocation recorded.

Also present in other papers' code

Of 14 samples lifted from this paper's linked repositories, 14 carry a function identity and 0 do not. Of the 14 distinct functions with an identity, 0 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 0 later) and 14 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.

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

A good motion retargeting cannot be reached without reasonable consideration of source-target differences on both the skeleton and shape geometry levels. In this work, we propose a novel Residual RETargeting network (R 2 ET) structure, which relies on two neural modification modules, to adjust the source motions to fit the target skeletons and shapes progressively. In particular, a skeleton-aware module is introduced to preserve the source motion semantics. A shape-aware module is designed to perceive the geometries of target characters to reduce interpenetration and contactmissing. Driven by our explored distance-based losses that explicitly model the motion semantics and geometry, these two modules can learn residual motion modifications on the source motion to generate plausible retargeted motion in a single inference without post-processing. To balance these two modifications, we further present a balancing gate to conduct linear interpolation between them. Extensive experiments on the public dataset Mixamo demonstrate that our R 2 ET achieves the state-of-the-art performance, and provides a good balance between the preservation of motion semantics as well as the attenuation of interpenetration and contact-missing. Code is available at https://github.com/Kebii/R2ET.

For agents

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

get_harvested_code_for_paper("2303.08658")
get_code_for_paper("2303.08658")
have("2303.08658")

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

curl https://syntology.ai/api/ran/2303.08658.json

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

[![Syntology run record](https://syntology.ai/api/ran/2303.08658.svg)](https://syntology.ai/paper/2303.08658)

Connect an agent — have() is free.