SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2407.16658 · ECCV · 2024

EgoCVR: An Egocentric Benchmark for Fine-Grained Composed Video Retrieval

Zeynep Akata, Mariana-Iuliana Georgescu, Thomas Hummel, Shyamgopal Karthik

arXiv · PDF · Open in the Atlas

Code that ran

10 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; 3 honoured the contract we drafted, 0 violated it and 0 have no contract verdict; 7 failures of our instrument; 4 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/2407.16658.json · badge: /api/ran/2407.16658.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
explainableml/egocvr canonical (classifier label)
classifier basis: census
9 ran · 4 no recorded run
ExplainableML/EgoCVR community
Papers with Code archive
1 ran · 0 no recorded run
FunctionStatusWhere it lives
get_middle_frame Ran
ran, honoured the contract we drafted
explainableml/egocvr/model/utils.py
code served (permissive licence) · get_code("fab5c9bd5d746577")
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 Image. 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 tempfile
import cv2
import numpy as np
from pathlib import Path

# Create a temporary video file with a few frames
temp_dir = tempfile.mkdtemp()
video_path = Path(temp_dir) / "test_video.mp4"

# Create a simple video file using cv2
fourcc = cv2.VideoWriter_fourcc(*'mp4v')
out = cv2.VideoWriter(str(video_path), fourcc, 30.0, (640, 480))

# Write 10 frames to the video
for i in range(10):
    frame = np.random.randint(0, 255, (480, 640, 3), dtype=np.uint8)
    out.write(frame)

out.release()

ARGS = {
    'reference_vid_pth': str(video_path)
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
PIL, cv2, numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Image — a Image output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a Image output the recorded ndim 3 and dtype family int 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
interpolate_pos_embed Ran
ran; our contract draft was wrong, not the code
explainableml/egocvr/model/blip/vit.py
code served (permissive licence) · get_code("c6ec173f19f5c34d")
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 Tensor, ndim 3, dtype family float, shape (1, 17, 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
import torch.nn as nn

torch.manual_seed(0)

class PatchEmbed(nn.Module):
    def __init__(self, num_patches):
        super().__init__()
        self.num_patches = num_patches

class VisualEncoder(nn.Module):
    def __init__(self, num_patches, pos_embed_shape):
        super().__init__()
        self.patch_embed = PatchEmbed(num_patches)
        self.pos_embed = torch.randn(pos_embed_shape)

# Create a visual encoder with 16 patches (4x4 grid)
# pos_embed shape: (1, num_patches + num_extra_tokens, embedding_dim)
num_patches = 16
num_extra_tokens = 1
embedding_size = 4
visual_encoder = VisualEncoder(
    num_patches=num_patches,
    pos_embed_shape=(1, num_patches + num_extra_tokens, embedding_size)
)

# Create pos_embed_checkpoint with original size 2x2 (4 patches) + 1 extra token
# This will trigger interpolation from 2x2 to 4x4
orig_size = 2
orig_num_patches = orig_size * orig_size
pos_embed_checkpoint = torch.randn(1, orig_num_patches + num_extra_tokens, embedding_size)

ARGS = {
    "pos_embed_checkpoint": pos_embed_checkpoint,
    "visual_encoder": visual_encoder,
}
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 Tensor, ndim 3, dtype family float, shape (1, 17, 4) — shape (1, 17, 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
nearest_neighbors Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
ExplainableML/EgoCVR/egocvr_retrieval.py
code served (permissive licence) · get_code("32b5a16451d8c9ac")
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 tuple, ndim 3, 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
import torch.nn.functional as F

torch.manual_seed(0)

# Create candidate embeddings: shape (num_candidates, embedding_dim)
candidate_embeddings = torch.randn(5, 4)

# Create query embedding: shape (embedding_dim,) - will be unsqueezed to (1, embedding_dim)
query = torch.randn(4)

# Number of nearest neighbors to retrieve
k = 3

# Normalization flag
normalize = True

# Return distances flag
return_distances = True

ARGS = {
    "candidate_embeddings": candidate_embeddings,
    "query": query,
    "k": k,
    "normalize": normalize,
    "return_distances": return_distances,
}
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 tuple, ndim 3, 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
parse_clip_name Ran
ran; our contract draft was wrong, not the code
explainableml/egocvr/egocvr_download_and_extract.py
code served (permissive licence) · get_code("d1bab1a6e28b79d1")
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. 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 re

# The function parse_clip_name takes a single parameter: clip_name (str)
# It parses a clip_name string in the format:
# <video_uid>_<start_token>_<end_token>
# where video_uid is a UUID in the format 8-4-4-4-12 hex digits separated by hyphens

# Valid example clip_name:
# "550e8400-e29b-41d4-a716-446655440000_971_243-2"
# This has:
# - video_uid: 550e8400-e29b-41d4-a716-446655440000 (valid UUID)
# - start_token: 971 (plain integer, parses to 971.0)
# - end_token: 243-2 (dash format, parses to 243.2)

clip_name_value = "550e8400-e29b-41d4-a716-446655440000_971_243-2"

ARGS = {
    "clip_name": clip_name_value
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — 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
parse_timestamp Ran
ran, honoured the contract we drafted
explainableml/egocvr/egocvr_download_and_extract.py
code served (permissive licence) · get_code("cbc6ea444c031c11")
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 float, 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 torch

ARGS = {
    "token": "243-2"
}
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 float, 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
pre_caption Ran
ran; our contract draft was wrong, not the code
explainableml/egocvr/model/utils.py
pointer only (licence: MIT); the stored copy get_code serves, from amitakamath/hard_positives, is pointer only (licence: NONE) · get_code("0d760ca65749dc14")
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 amitakamath/hard_positives, 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.

remove_non_digits Ran
ran, honoured the contract we drafted
explainableml/egocvr/model/utils.py
code served (permissive licence) · get_code("7cca1d789edd477c")
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 int, ndim 0, dtype family int, 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 re

def remove_non_digits(string, sub: str=''):
    return int(re.sub('\\D', sub, string))

ARGS = {
    'string': 'abc123def456ghi',
    'sub': ''
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type int, ndim 0, dtype family int, 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
sim_matrix Ran
ran; our contract draft was wrong, not the code
explainableml/egocvr/model/egovlpv2/model.py
pointer only (licence: MIT); the stored copy get_code serves, from showlab/egovlp, is pointer only (licence: NONE) · get_code("185c7b880ca6135e")
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 showlab/egovlp, 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.

sim_matrix_batch_val Ran
ran; our contract draft was wrong, not the code
explainableml/egocvr/model/egovlpv2/model.py
code served (permissive licence) · get_code("eb002dfbd2042d75")
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, 5) 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 two batch tensors for similarity matrix computation
# The function expects 3D tensors: (batch_size, seq_len, embedding_dim)
# Using distinct sizes: batch=2, seq_len_a=3, seq_len_b=5, embedding_dim=4

a = torch.randn(2, 3, 4)  # batch_size=2, seq_len=3, embedding_dim=4
b = torch.randn(2, 5, 4)  # batch_size=2, seq_len=5, embedding_dim=4

ARGS = {
    "a": a,
    "b": b,
    "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 3, dtype family float, shape (2, 3, 5) — shape (2, 3, 5) 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
temporal_sample Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
explainableml/egocvr/model/models.py
code served (permissive licence) · get_code("8558a76e8b3fa190")
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, shape (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 numpy as np

torch.manual_seed(0)

# The function takes embeddings (a tensor) and num_frames (an integer)
# It checks ndim, len(), and supports indexing and mean operations

# Case 1: 1D embedding
embeddings_1d = torch.randn(4)

# Case 2: 2D embedding (multiple frames)
embeddings_2d = torch.randn(5, 3)

# num_frames parameter
num_frames = 2

# Build ARGS with a 2D embedding to test the main logic
ARGS = {
    "embeddings": embeddings_2d,
    "num_frames": num_frames
}
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 float, shape (3,) — shape (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
dump_embeddings Not yet run
unverified (harvested, no recorded run)
explainableml/egocvr/egocvr_retrieval.py
code served (permissive licence) · get_code("8587e9f0fd81f3fa")
How it ran

no invocation recorded.

extract_clip Not yet run
unverified (harvested, no recorded run)
explainableml/egocvr/egocvr_download_and_extract.py
code served (permissive licence) · get_code("8de3d9d9a8d90470")
How it ran

no invocation recorded.

forward_blip_visual Not yet run
unverified (harvested, no recorded run)
explainableml/egocvr/model/models.py
code served (permissive licence) · get_code("2e6e17098f068d00")
How it ran

no invocation recorded.

load_embeddings Not yet run
unverified (harvested, no recorded run)
explainableml/egocvr/egocvr_retrieval.py
code served (permissive licence) · get_code("0bdbe9b1b43f5498")
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, 4 are also present in other papers whose code we hold (4 with at least one paper dated earlier, 0 the same month, 2 later) and 10 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
interpolate_pos_embed 52 other papers
41 dated earlier, 9 later, 2 same year, order not determinable
1905.12794 (2019-05)
Fashion IQ: A New Dataset Towards Retrieving Images by Natural Language Feedback
2111.09452 (2021-11)
Open Vocabulary Object Detection with Pseudo Bounding-Box Labels
2205.11169 (2022-05)
PEVL: Position-enhanced Pre-training and Prompt Tuning for Vision-language Models
and 49 more not named here
sim_matrix 8 other papers
8 dated earlier
2104.00650 (2021-04)
Frozen in Time: A Joint Video and Image Encoder for End-to-End Retrieval
2022.findings-emnlp.24 (2022 (year only))
AssistSR: Task-oriented Video Segment Retrieval for Personal AI Assistant
2206.01670 (2022-06)
Egocentric Video-Language Pretraining
and 5 more not named here
pre_caption 5 other papers
4 dated earlier, 1 later
2210.01936 (2022-10)
When and why vision-language models behave like bags-of-words, and what to do about it?
2303.10323 (2023-03)
Dynamic Graph Enhanced Contrastive Learning for Chest X-ray Report Generation
2310.07972 (2023-10)
Interpretable Diffusion via Information Decomposition
and 2 more not named here
sim_matrix_batch_val 1 other paper
1 dated earlier
2311.18259 (2023-11)
Ego-Exo4D: Understanding Skilled Human Activity from First- and Third-Person Perspectives

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

In Composed Video Retrieval, a video and a textual description which modifies the video content are provided as inputs to the model. The aim is to retrieve the relevant video with the modified content from a database of videos. In this challenging task, the first step is to acquire large-scale training datasets and collect high-quality benchmarks for evaluation. In this work, we introduce EgoCVR, a new evaluation benchmark for fine-grained Composed Video Retrieval using large-scale egocentric video datasets. EgoCVR consists of 2,295 queries that specifically focus on high-quality temporal video understanding. We find that existing Composed Video Retrieval frameworks do not achieve the necessary high-quality temporal video understanding for this task. To address this shortcoming, we adapt a simple training-free method, propose a generic re-ranking framework for Composed Video Retrieval, and demonstrate that this achieves strong results on EgoCVR. Our code and benchmark are freely available at https://github.com/ExplainableML/EgoCVR.

For agents

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

get_harvested_code_for_paper("2407.16658")
get_code_for_paper("2407.16658")
have("2407.16658")

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

curl https://syntology.ai/api/ran/2407.16658.json

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

[![Syntology run record](https://syntology.ai/api/ran/2407.16658.svg)](https://syntology.ai/paper/2407.16658)

Connect an agent — have() is free.