SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2303.12930 · CVPR · 2023

Dense-Localizing Audio-Visual Events in Untrimmed Videos: A Large-Scale Benchmark and Baseline

Runmin Cong, Feng Zheng, Teng Wang, Jinming Duan, Tiantian Geng

arXiv · PDF · Open in the Atlas

Code that ran

11 of 13 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; 1 honoured the contract we drafted, 0 violated it and 1 has no contract verdict; 9 failures of our instrument; 2 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/2303.12930.json · badge: /api/ran/2303.12930.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
ttgeng233/UnAV canonical (classifier label)
classifier basis not recorded on the link
11 ran · 2 no recorded run
FunctionStatusWhere it lives
ctr_diou_loss_1d Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
ttgeng233/UnAV/libs/modeling/losses.py
code served (permissive licence) · get_code("be1ca95e264f135e")
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 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 torch

torch.manual_seed(0)

# Create small 1D offset tensors
# Shape: (N, 2) where N=2 is batch size, 2 represents left and right offsets
input_offsets = torch.tensor([
    [0.5, 1.5],
    [1.0, 2.0]
], dtype=torch.float32)

target_offsets = torch.tensor([
    [0.7, 1.3],
    [0.9, 2.1]
], dtype=torch.float32)

ARGS = {
    'input_offsets': input_offsets,
    'target_offsets': target_offsets,
    'reduction': 'none',
    'class_aware': False,
    'eps': 1e-08
}
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 1, dtype family float — no output shape is stated: observe_signature.py could not observe it (import:OSError)
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
drop_path Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
ttgeng233/UnAV/libs/modeling/blocks.py
pointer only (licence: MIT); the stored copy get_code serves, from f3set/f3set, is pointer only (licence: NONE) · get_code("a34c005ba2203f35")
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 f3set/f3set, 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.

get_sinusoid_encoding Ran
ran, honoured the contract we drafted
ttgeng233/UnAV/libs/modeling/blocks.py
pointer only (licence: MIT); the stored copy get_code serves, from f3set/f3set, is pointer only (licence: NONE) · get_code("3531df7b0c9b0791")
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 f3set/f3set, 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.

register_dataset Ran
ran; our contract draft was wrong, not the code
ttgeng233/UnAV/libs/datasets/datasets.py
pointer only (licence: MIT); the stored copy get_code serves, from zzhhfut/ccnet-aaai2025, is pointer only (licence: NONE) · get_code("2ee0df0825339a6e")
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 zzhhfut/ccnet-aaai2025, 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.

register_dependency_block Ran
ran on a synthesized input
ttgeng233/UnAV/libs/modeling/models.py
code served (permissive licence) · get_code("d8a13c98d784c121")
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 function. 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)

# Define a simple class to be registered as a dependency block
class SimpleDependencyBlock(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(4, 3)
    
    def forward(self, x):
        return self.linear(x)

# The function register_dependency_block takes a single parameter: name (str)
# It returns a decorator that registers a class in the dependency_blocks dict

ARGS = {
    "name": "simple_block"
}
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 function — a function output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a function 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
register_generator Ran
ran; our contract draft was wrong, not the code
ttgeng233/UnAV/libs/datasets/datasets.py
pointer only (licence: MIT); the stored copy get_code serves, from zzhhfut/ccnet-aaai2025, is pointer only (licence: NONE) · get_code("3db1c5ab5f3f0198")
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 zzhhfut/ccnet-aaai2025, 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.

register_multimodal_backbone Ran
ran; our contract draft was wrong, not the code
ttgeng233/UnAV/libs/modeling/models.py
code served (permissive licence) · get_code("e85f9ec07f7d587b")
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 function. 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 register_multimodal_backbone takes a single parameter: name
# It returns a decorator function that registers a class in the multimodal_backbones dict
# To set up valid arguments, we need to provide a string name

class DummyMultimodalBackbone(nn.Module):
    """A simple stand-in multimodal backbone"""
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(4, 8)
    
    def forward(self, x):
        return self.linear(x)

# The function is a decorator factory, so we call it with a name string
# Then apply it to a class
ARGS = {
    "name": "dummy_backbone_v1"
}
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 function — a function output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a function 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
register_multimodal_meta_arch Ran
ran; our contract draft was wrong, not the code
ttgeng233/UnAV/libs/modeling/models.py
code served (permissive licence) · get_code("0fc8e74a8c092f28")
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 function. 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 register_multimodal_meta_arch takes a single parameter: name (a string)
# It returns a decorator function that registers a class in the multimodal_meta_archs dict
# The decorator just stores the class and returns it unchanged

# Create a simple test class to register
class SimpleMultimodalArch(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(4, 3)
    
    def forward(self, x):
        return self.linear(x)

# The only parameter is 'name', which should be a string identifier
ARGS = {
    "name": "simple_multimodal_model"
}
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 function — a function output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a function 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
sigmoid_focal_loss Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
ttgeng233/UnAV/libs/modeling/losses.py
pointer only (licence: MIT); the stored copy get_code serves, from yangle15/DyFADet-pytorch, is pointer only (licence: NONE) · get_code("54293ea22bd4f952")
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 yangle15/DyFADet-pytorch, 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.

trivial_batch_collator Ran
ran; our contract draft was wrong, not the code
ttgeng233/UnAV/libs/datasets/data_utils.py
pointer only (licence: MIT); the stored copy get_code serves, from zzhhfut/ccnet-aaai2025, is pointer only (licence: NONE) · get_code("0c250bf03b8c9311")
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 zzhhfut/ccnet-aaai2025, 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.

truncate_feats Ran
ran; our contract draft was wrong, not the code
ttgeng233/UnAV/libs/datasets/data_utils.py
pointer only (licence: MIT); the stored copy get_code serves, from zzhhfut/ccnet-aaai2025, is pointer only (licence: NONE) · get_code("0187c7d2e60afe31")
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 zzhhfut/ccnet-aaai2025, 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.

load_config Not yet run
unverified (harvested, no recorded run)
ttgeng233/UnAV/libs/core/config.py
code served (permissive licence) · get_code("157c518e791fca20")
How it ran

no invocation recorded.

make_dataset Not yet run
unverified (harvested, no recorded run)
ttgeng233/UnAV/libs/datasets/datasets.py
pointer only (licence: MIT); the stored copy get_code serves, from zzhhfut/ccnet-aaai2025, is pointer only (licence: NONE) · get_code("ffebfbb8094c0dcc")
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 zzhhfut/ccnet-aaai2025, 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. No invocation recorded.

Also present in other papers' code

Of 13 samples lifted from this paper's linked repositories, 13 carry a function identity and 0 do not. Of the 13 distinct functions with an identity, 12 are also present in other papers whose code we hold (3 with at least one paper dated earlier, 8 the same month, 12 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
drop_path 11 other papers
1 dated earlier, 1 the same month, 9 later
2104.14222 (2021-04)
Privacy-Preserving Portrait Matting
2303.07347 (2023-03)
TriDet: Temporal Action Detection with Relative Boundary Modeling
2404.02257 (2024-04)
SnAG: Scalable and Accurate Video Grounding
and 8 more not named here
get_sinusoid_encoding 11 other papers
1 dated earlier, 1 the same month, 9 later
2104.14222 (2021-04)
Privacy-Preserving Portrait Matting
2303.07347 (2023-03)
TriDet: Temporal Action Detection with Relative Boundary Modeling
2404.02257 (2024-04)
SnAG: Scalable and Accurate Video Grounding
and 8 more not named here
trivial_batch_collator 9 other papers
2 dated earlier, 1 the same month, 6 later
2208.05647 (2022-08)
PPMN: Pixel-Phrase Matching Network for One-Stage Panoptic Narrative Grounding
2211.05272 (2022-11)
GAPartNet: Cross-Category Domain-Generalizable Object Perception and Manipulation via Generalizable and Actionable Parts
2303.07347 (2023-03)
TriDet: Temporal Action Detection with Relative Boundary Modeling
and 6 more not named here
load_config 8 other papers
1 the same month, 7 later
2303.07347 (2023-03)
TriDet: Temporal Action Detection with Relative Boundary Modeling
2404.02257 (2024-04)
SnAG: Scalable and Accurate Video Grounding
2406.15778 (2024-06)
ObjectNLQ @ Ego4D Episodic Memory Challenge 2024
and 5 more not named here
register_generator 8 other papers
1 the same month, 7 later
2303.07347 (2023-03)
TriDet: Temporal Action Detection with Relative Boundary Modeling
2404.02257 (2024-04)
SnAG: Scalable and Accurate Video Grounding
2404.03179 (2024-04)
UniAV: Unified Audio-Visual Perception for Multi-Task Video Event Localization
and 5 more not named here
sigmoid_focal_loss 7 other papers
1 the same month, 6 later
2303.07347 (2023-03)
TriDet: Temporal Action Detection with Relative Boundary Modeling
2312.06505 (2023-12)
Grounded Question-Answering in Long Egocentric Videos
2404.02257 (2024-04)
SnAG: Scalable and Accurate Video Grounding
and 4 more not named here
make_dataset 4 other papers
1 the same month, 3 later
2303.07347 (2023-03)
TriDet: Temporal Action Detection with Relative Boundary Modeling
2404.02257 (2024-04)
SnAG: Scalable and Accurate Video Grounding
2404.03179 (2024-04)
UniAV: Unified Audio-Visual Perception for Multi-Task Video Event Localization
and 1 more not named here
register_dataset 4 other papers
1 the same month, 3 later
2303.07347 (2023-03)
TriDet: Temporal Action Detection with Relative Boundary Modeling
2404.02257 (2024-04)
SnAG: Scalable and Accurate Video Grounding
2404.03179 (2024-04)
UniAV: Unified Audio-Visual Perception for Multi-Task Video Event Localization
and 1 more not named here
ctr_diou_loss_1d 2 other papers
2 later
2404.03179 (2024-04)
UniAV: Unified Audio-Visual Perception for Multi-Task Video Event Localization
2412.12628 (2024-12)
Dense Audio-Visual Event Localization under Cross-Modal Consistency and Multi-Temporal Granularity Collaboration
register_multimodal_backbone 1 other paper
1 later
2404.03179 (2024-04)
UniAV: Unified Audio-Visual Perception for Multi-Task Video Event Localization
register_multimodal_meta_arch 1 other paper
1 later
2404.03179 (2024-04)
UniAV: Unified Audio-Visual Perception for Multi-Task Video Event Localization
truncate_feats 1 other paper
1 later
2412.12628 (2024-12)
Dense Audio-Visual Event Localization under Cross-Modal Consistency and Multi-Temporal Granularity Collaboration

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

Existing audio-visual event localization (AVE) handles manually trimmed videos with only a single instance in each of them. However, this setting is unrealistic as natural videos often contain numerous audio-visual events with different categories. To better adapt to real-life applications, in this paper we focus on the task of denselocalizing audio-visual events, which aims to jointly localize and recognize all audio-visual events occurring in an untrimmed video. The problem is challenging as it requires fine-grained audio-visual scene and context understanding. To tackle this problem, we introduce the first Untrimmed Audio-Visual (UnAV-100) dataset, which contains 10K untrimmed videos with over 30K audio-visual events. Each video has 2.8 audio-visual events on average, and the events are usually related to each other and might co-occur as in real-life scenes. Next, we formulate the task using a new learning-based framework, which is capable of fully integrating audio and visual modalities to localize audio-visual events with various lengths and capture dependencies between them in a single pass. Extensive experiments demonstrate the effectiveness of our method as well as the significance of multi-scale cross-modal perception and dependency modeling for this task. The dataset and code are available at https://unav100.github.io.

For agents

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

get_harvested_code_for_paper("2303.12930")
get_code_for_paper("2303.12930")
have("2303.12930")

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

curl https://syntology.ai/api/ran/2303.12930.json

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

[![Syntology run record](https://syntology.ai/api/ran/2303.12930.svg)](https://syntology.ai/paper/2303.12930)

Connect an agent — have() is free.