SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2203.07918 · CVPR · 2022

GPV-Pose: Category-level Object Pose Estimation via Geometry-guided Point-wise Voting

arXiv · PDF · Open in the Atlas

Code that ran

6 of 9 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; 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/2203.07918.json · badge: /api/ran/2203.07918.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
lolrudy/gpv_pose canonical (classifier label)
classifier basis not recorded on the link
6 ran · 3 no recorded run
FunctionStatusWhere it lives
control_loss Ran
ran on a synthesized input
lolrudy/gpv_pose/engine/organize_loss.py
code served (permissive licence) · get_code("c26e5139544812be")
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 torch
import torch.nn as nn

torch.manual_seed(0)

ARGS = {
    'Train_stage': 'PoseNet_only'
}
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 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
defor_2D Ran
ran on a synthesized input
lolrudy/gpv_pose/datasets/data_augmentation.py
code served (permissive licence) · get_code("1907108fce7bbe9a")
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 (256, 256) 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

np.random.seed(0)

# Create a 256x256 binary mask
roi_mask = np.random.randint(0, 2, size=(256, 256)).astype(np.float32)

ARGS = {
    "roi_mask": roi_mask,
    "rand_r": 2,
    "rand_pro": 0.3
}
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 2, dtype family float, shape (256, 256) — shape (256, 256) 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
defor_3D_bb Ran
ran on a synthesized input
lolrudy/gpv_pose/datasets/data_augmentation.py
code served (permissive licence) · get_code("479152e874348de6")
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 torch

torch.manual_seed(0)

# Construct arguments for defor_3D_bb function
# pc: point cloud, shape (N, 3) where N is number of points
pc = torch.randn(2, 3)

# R: rotation matrix, shape (3, 3)
R = torch.eye(3)

# t: translation vector, shape (3,)
t = torch.randn(3)

# s: scale vector, shape (3,)
s = torch.tensor([1.0, 1.0, 1.0])

# sym: symmetry flag, list/tuple with at least one element
# sym[0] == 1 means symmetric case, otherwise asymmetric
sym = [1]

# aug_bb: augmentation bounding box, list/tuple of 3 scale factors (ex, ey, ez)
aug_bb = [1.2, 0.9, 1.1]

ARGS = {
    'pc': pc,
    'R': R,
    't': t,
    's': s,
    'sym': sym,
    'aug_bb': aug_bb
}
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 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
defor_3D_bc Ran
ran on a synthesized input
lolrudy/gpv_pose/datasets/data_augmentation.py
code served (permissive licence) · get_code("d00ebddb21609c8d")
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 torch

torch.manual_seed(0)

# pc: point cloud, shape (N, 3)
pc = torch.randn(4, 3)

# R: rotation matrix, shape (3, 3)
R = torch.eye(3)

# t: translation vector, shape (3,)
t = torch.randn(3)

# s: scale vector with at least 2 elements (s[1] is accessed)
s = torch.tensor([1.0, 2.0])

# model_point: model points, shape (M, 3)
model_point = torch.randn(5, 3)

# nocs_scale: scalar for scaling
nocs_scale = 0.5

ARGS = {
    "pc": pc,
    "R": R,
    "t": t,
    "s": s,
    "model_point": model_point,
    "nocs_scale": nocs_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 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
expected_wrt_prob Ran
ran on a synthesized input
lolrudy/gpv_pose/nnutils/layers.py
code served (permissive licence) · get_code("6709651c2a3bfacd")
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, 3, 5, 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 torch

torch.manual_seed(0)

# Construct arguments for expected_wrt_prob
# value: (N, C, D, H, W) - a tensor with values
# prob: (N, 1, D + 1, H, W) - probabilities tensor
# miss: float - scalar for missing/background

N, C, D, H, W = 2, 3, 4, 5, 6

# value tensor: (N=2, C=3, D=4, H=5, W=6)
value = torch.randn(N, C, D, H, W)

# prob tensor: (N=2, 1, D+1=5, H=5, W=6)
# This has D+1 in the depth dimension for foreground + background split
prob = torch.softmax(torch.randn(N, 1, D + 1, H, W), dim=2)

# miss: scalar float
miss = 0.1

ARGS = {
    "value": value,
    "prob": prob,
    "miss": miss,
}
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, 3, 5, 6) — shape (2, 3, 5, 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
laplacian_cot Ran
ran on a synthesized input
lolrudy/gpv_pose/nnutils/laplacian_loss.py
code served (permissive licence) · get_code("715f90a0136b95bb")
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 torch

torch.manual_seed(0)

class MockMeshes:
    """Minimal Meshes object that provides the interface used by laplacian_cot"""
    def __init__(self, verts, faces):
        self._verts = verts
        self._faces = faces
        self.device = verts.device
    
    def verts_packed(self):
        """Return packed vertices"""
        return self._verts
    
    def faces_packed(self):
        """Return packed faces (indices into vertices)"""
        return self._faces

# Create a simple mesh: 4 vertices forming a tetrahedron
verts = torch.tensor([
    [0.0, 0.0, 0.0],
    [1.0, 0.0, 0.0],
    [0.0, 1.0, 0.0],
    [0.0, 0.0, 1.0]
], dtype=torch.float32)

# Create faces (triangles) - each row is 3 vertex indices
faces = torch.tensor([
    [0, 1, 2],
    [0, 1, 3],
    [0, 2, 3],
    [1, 2, 3]
], dtype=torch.long)

# Create the meshes object
meshes = MockMeshes(verts, faces)

ARGS = {
    'meshes': meshes
}
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 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
expected_wrt_occupancy Not yet run
unverified (harvested, no recorded run)
lolrudy/gpv_pose/nnutils/layers.py
code served (permissive licence) · get_code("b7549cd87fb0c3ec")
How it ran

no invocation recorded.

mesh_laplacian_smoothing Not yet run
unverified (harvested, no recorded run)
lolrudy/gpv_pose/nnutils/laplacian_loss.py
code served (permissive licence) · get_code("ac4d8b3cbad04fcc")
How it ran

no invocation recorded.

occupancy_to_prob Not yet run
unverified (harvested, no recorded run)
lolrudy/gpv_pose/nnutils/layers.py
code served (permissive licence) · get_code("ded17a074ea35ec9")
How it ran

no invocation recorded.

Also present in other papers' code

Of 9 samples lifted from this paper's linked repositories, 9 carry a function identity and 0 do not. Of the 9 distinct functions with an identity, 8 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 8 later) and 1 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
control_loss 1 other paper
1 later
2303.15743 (2023-03)
HS-Pose: Hybrid Scope Feature Extraction for Category-level Object Pose Estimation
defor_2D 1 other paper
1 later
2303.15743 (2023-03)
HS-Pose: Hybrid Scope Feature Extraction for Category-level Object Pose Estimation
defor_3D_bb 1 other paper
1 later
2303.15743 (2023-03)
HS-Pose: Hybrid Scope Feature Extraction for Category-level Object Pose Estimation
expected_wrt_occupancy 1 other paper
1 later
2303.15743 (2023-03)
HS-Pose: Hybrid Scope Feature Extraction for Category-level Object Pose Estimation
expected_wrt_prob 1 other paper
1 later
2303.15743 (2023-03)
HS-Pose: Hybrid Scope Feature Extraction for Category-level Object Pose Estimation
laplacian_cot 1 other paper
1 later
2303.15743 (2023-03)
HS-Pose: Hybrid Scope Feature Extraction for Category-level Object Pose Estimation
mesh_laplacian_smoothing 1 other paper
1 later
2303.15743 (2023-03)
HS-Pose: Hybrid Scope Feature Extraction for Category-level Object Pose Estimation
occupancy_to_prob 1 other paper
1 later
2303.15743 (2023-03)
HS-Pose: Hybrid Scope Feature Extraction for Category-level Object Pose Estimation

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

While 6D object pose estimation has recently made a huge leap forward, most methods can still only handle a single or a handful of different objects, which limits their applications. To circumvent this problem, category-level object pose estimation has recently been revamped, which aims at predicting the 6D pose as well as the 3D metric size for previously unseen instances from a given set of object classes. This is, however, a much more challenging task due to severe intra-class shape variations. To address this issue, we propose GPV-Pose, a novel framework for robust category-level pose estimation, harnessing geometric insights to enhance the learning of category-level pose-sensitive features. First, we introduce a decoupled confidence-driven rotation representation, which allows geometry-aware recovery of the associated rotation matrix. Second, we propose a novel geometry-guided point-wise voting paradigm for robust retrieval of the 3D object bounding box. Finally, leveraging these different output streams, we can enforce several geometric consistency terms, further increasing performance, especially for non-symmetric categories. GPV-Pose produces superior results to state-of-the-art competitors on common public benchmarks, whilst almost achieving real-time inference speed at 20 FPS.

For agents

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

get_harvested_code_for_paper("2203.07918")
get_code_for_paper("2203.07918")
have("2203.07918")

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

curl https://syntology.ai/api/ran/2203.07918.json

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

[![Syntology run record](https://syntology.ai/api/ran/2203.07918.svg)](https://syntology.ai/paper/2203.07918)

Connect an agent — have() is free.