SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2203.05203 · ECCV · 2022

MORE: Multi-Order RElation Mining for Dense Captioning in 3D Scenes

Lin Ma, Yu-Gang Jiang, Yang Jiao, Jingjing Chen, Zequn Jie, Shaoxiang Chen

arXiv · PDF · Open in the Atlas

Code that ran

8 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; 3 honoured the contract we drafted, 0 violated it and 4 have no contract verdict; 1 failure of our instrument; 1 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/2203.05203.json · badge: /api/ran/2203.05203.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
SxJyJay/MORE canonical (classifier label)
classifier basis not recorded on the link
8 ran · 1 no recorded run
FunctionStatusWhere it lives
check_candidates Ran
ran on a synthesized input
SxJyJay/MORE/lib/eval_helper.py
code served (permissive licence) · get_code("4795981076ddc399")
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 dict. 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 numpy as np

torch.manual_seed(0)

# The function check_candidates takes two parameters:
# - corpus: a dict with keys that will be checked
# - candidates: a dict that may be missing some keys from corpus
# 
# The function:
# 1. Gets keys from both dicts
# 2. Finds keys in corpus that are not in candidates
# 3. For missing keys, adds them to candidates with value [placeholder]
# 4. Returns the modified candidates dict

# Create a corpus dict with several keys
corpus = {
    'doc1': 'some content',
    'doc2': 'more content',
    'doc3': 'even more content',
    'doc4': 'last content'
}

# Create a candidates dict that is missing some keys from corpus
candidates = {
    'doc1': ['candidate1', 'candidate2'],
    'doc3': ['candidate3']
    # Missing: doc2 and doc4
}

ARGS = {
    'corpus': corpus,
    'candidates': candidates
}
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 dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict 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
compute_cap_loss Ran
ran on a synthesized input
SxJyJay/MORE/lib/loss_helper_pretrained.py
code served (permissive licence) · get_code("3e1db4a4fe30db4a")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned tuple, ndim 1, dtype family float; no output shape is stated. 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 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 torch
import torch.nn as nn

torch.manual_seed(0)

# Create small tensors with distinct dimensions
batch_size = 2
num_words = 4
num_vocabs = 5

# lang_cap: (batch_size, num_words-1, num_vocabs) - predicted captions
lang_cap = torch.randn(batch_size, num_words - 1, num_vocabs)

# lang_ids: (batch_size, num_words) - target caption ids
lang_ids = torch.randint(0, num_vocabs, (batch_size, num_words))
lang_ids[:, 0] = 0  # ensure first token is special token

# lang_len: (batch_size,) - length of each caption
lang_len = torch.tensor([num_words, num_words])

# good_bbox_masks: (batch_size,) - boolean mask for valid bounding boxes
good_bbox_masks = torch.tensor([True, False])

# Create data_dict for 'gt' mode
data_dict_gt = {
    'lang_cap': lang_cap,
    'lang_ids': lang_ids,
    'lang_len': lang_len,
    'good_bbox_masks': good_bbox_masks,
}

# Create data_dict for 'votenet' mode (needs good_bbox_masks)
data_dict_votenet = {
    'lang_cap': lang_cap,
    'lang_ids': lang_ids,
    'lang_len': lang_len,
    'good_bbox_masks': good_bbox_masks,
}

ARGS = {
    'data_dict': data_dict_gt,
    'mode': 'gt',
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple, ndim 1, dtype family float — a tuple output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
create_enet Ran
ran; our contract draft was wrong, not the code
SxJyJay/MORE/lib/enet.py
pointer only (licence: Apache-2.0); the stored copy get_code serves, from matthewdm0816/bridgeqa, is pointer only (licence: NOASSERTION) · get_code("d209e4ea51338ac8")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (Apache-2.0), but get_code serves the stored copy, first harvested from matthewdm0816/bridgeqa, and GitHub could not classify the licence of that repository (NOASSERTION). 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.

decode_caption Ran
ran on a synthesized input
SxJyJay/MORE/lib/eval_helper.py
code served (permissive licence) · get_code("239555e6188daee7")
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 str. 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)

# raw_caption: a tensor of token indices that will be iterated over
# Each element needs .item() to convert to Python int
raw_caption = torch.tensor([1, 2, 3, 4])

# idx2word: a dict mapping string indices to word tokens
# The function accesses idx2word[str(token_idx)] for each token
# We need to include 'eos' as a token to test the break condition
idx2word = {
    '0': 'sos',
    '1': 'hello',
    '2': 'world',
    '3': 'eos',
    '4': 'test'
}

ARGS = {
    'raw_caption': raw_caption,
    'idx2word': idx2word
}
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 str — a str output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a str 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
flip_axis_to_camera Ran
ran, honoured the contract we drafted
SxJyJay/MORE/lib/ap_helper.py
pointer only (licence: Apache-2.0); the stored copy get_code serves, from tingxueronghua/dpke, is pointer only (licence: NONE) · get_code("eddd23cbbb297813")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (Apache-2.0), but get_code serves the stored copy, first harvested from tingxueronghua/dpke, 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.

flip_axis_to_depth Ran
ran, honoured the contract we drafted
SxJyJay/MORE/lib/ap_helper.py
code served (permissive licence) · get_code("d638b0201b9a9ece")
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 (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 numpy as np

# The function takes a point cloud (pc) which is a numpy array
# It performs axis flipping and negation operations on it
# The array has shape (..., 3) where the last dimension contains [x, y, z] coordinates

# Create a small point cloud array with shape (2, 3)
# This represents 2 points in 3D space
pc = np.array([
    [1.0, 2.0, 3.0],
    [4.0, 5.0, 6.0]
], dtype=np.float32)

ARGS = {
    "pc": pc
}
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 (2, 3) — shape (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
radian_to_label Ran
ran on a synthesized input
SxJyJay/MORE/lib/loss_helper_pretrained.py
code served (permissive licence) · get_code("79493094543ea699")
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 1, dtype family int, shape (2,) 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
import numpy as np

torch.manual_seed(0)

# Create a small batch of radian values
batch_size = 2
radians = torch.tensor([0.5, 1.2], dtype=torch.float32)

# num_bins parameter
num_bins = 6

# Construct ARGS dict with all parameters
ARGS = {
    'radians': radians,
    'num_bins': num_bins
}
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 Tensor, ndim 1, dtype family int, 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
softmax Ran
ran, honoured the contract we drafted
SxJyJay/MORE/lib/ap_helper.py
code served (permissive licence) · get_code("1a034a72c15f9903")
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 (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 numpy as np

# Construct a small numpy array for the softmax function
# The function operates on the last axis, so we'll create a 2D array
# to demonstrate typical usage
x = np.array([
    [1.0, 2.0, 3.0, 4.0],
    [2.0, 1.0, 4.0, 3.0]
], dtype=np.float32)

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 ndarray, 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
create_enet_for_3d Not yet run
unverified (harvested, no recorded run)
SxJyJay/MORE/lib/enet.py
pointer only (licence: Apache-2.0); the stored copy get_code serves, from matthewdm0816/bridgeqa, is pointer only (licence: NOASSERTION) · get_code("f9f6ab5de42cc2d4")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (Apache-2.0), but get_code serves the stored copy, first harvested from matthewdm0816/bridgeqa, and GitHub could not classify the licence of that repository (NOASSERTION). 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.

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, 5 are also present in other papers whose code we hold (3 with at least one paper dated earlier, 3 the same month, 5 later) and 4 are only here among the 52,977 papers whose code Syntology holds (as of 2026-09-26).

Only here means: among the 52,977 papers whose code Syntology holds (51,887 with lifted samples, 1,090 through their assemblies only), as of 2026-09-26, no other paper's lifted samples or current assembly bricks contain an identical function body after normalisation.

FunctionAlso present inEarliest-datedOthers named
flip_axis_to_camera 10 other papers
3 dated earlier, 1 the same month, 6 later
1904.09664 (2019-04)
Deep Hough Voting for 3D Object Detection in Point Clouds
1912.11803 (2019-12)
SESS: Self-Ensembling Semi-Supervised 3D Object Detection
2004.05679 (2020-04)
MLCVNet: Multi-Level Context VoteNet for 3D Object Detection
and 7 more not named here
softmax 9 other papers
3 dated earlier, 1 the same month, 5 later
1904.09664 (2019-04)
Deep Hough Voting for 3D Object Detection in Point Clouds
1912.11803 (2019-12)
SESS: Self-Ensembling Semi-Supervised 3D Object Detection
2004.05679 (2020-04)
MLCVNet: Multi-Level Context VoteNet for 3D Object Detection
and 6 more not named here
flip_axis_to_depth 7 other papers
2 dated earlier, 1 the same month, 4 later
1904.09664 (2019-04)
Deep Hough Voting for 3D Object Detection in Point Clouds
1912.11803 (2019-12)
SESS: Self-Ensembling Semi-Supervised 3D Object Detection
2203.09440 (2022-03)
TO-Scene: A Large-scale Dataset for Understanding 3D Tabletop Scenes
and 4 more not named here
create_enet 2 other papers
2 later
2204.06272 (2022-04)
3D-SPS: Single-Stage 3D Visual Grounding via Referred Point Progressive Selection
2402.15933 (2024-02)
Bridging the Gap between 2D and 3D Visual Question Answering: A Fusion Approach for 3D VQA
create_enet_for_3d 2 other papers
2 later
2204.06272 (2022-04)
3D-SPS: Single-Stage 3D Visual Grounding via Referred Point Progressive Selection
2402.15933 (2024-02)
Bridging the Gap between 2D and 3D Visual Question Answering: A Fusion Approach for 3D VQA

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

3D dense captioning is a recently-proposed novel task, where point clouds contain more geometric information than the 2D counterpart. However, it is also more challenging due to the higher complexity and wider variety of inter-object relations contained in point clouds. Existing methods only treat such relations as by-products of object feature learning in graphs without specifically encoding them, which leads to sub-optimal results. In this paper, aiming at improving 3D dense captioning via capturing and utilizing the complex relations in the 3D scene, we propose MORE, a Multi-Order RElation mining model, to support generating more descriptive and comprehensive captions. Technically, our MORE encodes object relations in a progressive manner since complex relations can be deduced from a limited number of basic ones. We first devise a novel Spatial Layout Graph Convolution (SLGC), which semantically encodes several first-order relations as edges of a graph constructed over 3D object proposals. Next, from the resulting graph, we further extract multiple triplets which encapsulate basic first-order relations as the basic unit, and construct several Object-centric Triplet Attention Graphs (OTAG) to infer multi-order relations for every target object. The updated node features from OTAG are aggregated and fed into the caption decoder to provide abundant relational cues, so that captions including diverse relations with context objects can be generated. Extensive experiments on the Scan2Cap dataset prove the effectiveness of our proposed MORE and its components, and we also outperform the current state-of-the-art method. Our code is available at https://github.com/SxJyJay/MORE.

For agents

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

get_harvested_code_for_paper("2203.05203")
get_code_for_paper("2203.05203")
have("2203.05203")

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

curl https://syntology.ai/api/ran/2203.05203.json

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

[![Syntology run record](https://syntology.ai/api/ran/2203.05203.svg)](https://syntology.ai/paper/2203.05203)

Connect an agent — have() is free.