SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2203.08481 · CVPR · 2022

Pseudo-Q: Generating Pseudo Language Queries for Visual Grounding

Gao Huang, Shiji Song, Haojun Jiang, Yuanze Lin, Dongchen Han

arXiv · PDF · Open in the Atlas

Code that ran

13 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; 0 honoured the contract we drafted, 0 violated it and 6 have no contract verdict; 7 failures of our instrument; 0 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/2203.08481.json · badge: /api/ran/2203.08481.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
LeapLabTHU/Pseudo-Q canonical (classifier label)
classifier basis not recorded on the link
13 ran · 0 no recorded run
FunctionStatusWhere it lives
build_vl_transformer Ran
ran; our contract draft was wrong, not the code
LeapLabTHU/Pseudo-Q/models/vl_transformer.py
pointer only (licence: Apache-2.0); the stored copy get_code serves, from liuting20/MaPPER, is pointer only (licence: NONE) · get_code("e0451d070cb0b43e")
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 liuting20/MaPPER, 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.

convert_examples_to_features Ran
ran; our contract draft was wrong, not the code
LeapLabTHU/Pseudo-Q/datasets/data_loader.py
pointer only (licence: Apache-2.0); the stored copy get_code serves, from Marloweeee/BARE, is pointer only (licence: NONE) · get_code("694e413f977737ba")
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 Marloweeee/BARE, 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.

crop Ran
ran; our contract draft was wrong, not the code
LeapLabTHU/Pseudo-Q/datasets/transforms.py
pointer only (licence: Apache-2.0); the stored copy get_code serves, from liuting20/MaPPER, is pointer only (licence: NONE) · get_code("f4ac915e5456c67f")
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 liuting20/MaPPER, 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.

filter_detect_cls Ran
ran on a synthesized input
LeapLabTHU/Pseudo-Q/pseudo_sample_generation/get_detection_results.py
code served (permissive licence) · get_code("8aec679ea5e75af5")
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 bool, 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

torch.manual_seed(0)

# The function filter_detect_cls takes a single parameter cls_name
# Looking at the function body, it simply returns True regardless of input
# cls_name appears to be a string representing a class name

ARGS = {
    "cls_name": "person"
}
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 bool, 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
filter_detect_cls Ran
ran on a synthesized input
LeapLabTHU/Pseudo-Q/pseudo_sample_generation/pseudo_query_generation.py
code served (permissive licence) · get_code("2e6c5c28154620d6")
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 bool, 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
import torch.nn as nn

torch.manual_seed(0)

# Based on the function signature and body, filter_detect_cls takes a single parameter:
# cls_name - a class name (string or identifier)
# The function body just returns True, so it doesn't actually use the parameter
# in any meaningful way, but we should still provide a valid string value for it.

ARGS = {
    "cls_name": "person"
}
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 bool, 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_image_blob Ran
ran on a synthesized input
LeapLabTHU/Pseudo-Q/pseudo_sample_generation/utils/extract_utils.py
code served (permissive licence) · get_code("63fc3fd8a7bfffff")
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
import cv2

torch.manual_seed(0)

# Create a small test image (height=480, width=640, channels=3)
# Using uint8 values typical of an image
im = np.random.randint(0, 256, size=(480, 640, 3), dtype=np.uint8)

# Create pixel_means as a sequence of 3 values (one per channel)
pixel_means = [103.53, 116.28, 123.675]

ARGS = {
    'im': im,
    'pixel_means': pixel_means,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
cv2, 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
im_list_to_blob Ran
ran on a synthesized input
LeapLabTHU/Pseudo-Q/pseudo_sample_generation/utils/extract_utils.py
code served (permissive licence) · get_code("d614ca29b05d68ec")
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 4, dtype family float, shape (2, 4, 6, 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 a list of images with different shapes
# Each image should be (height, width, 3) for BGR order
im1 = np.random.randn(4, 5, 3).astype(np.float32)
im2 = np.random.randn(3, 6, 3).astype(np.float32)

ims = [im1, im2]

ARGS = {
    "ims": ims
}
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 4, dtype family float, shape (2, 4, 6, 3) — shape (2, 4, 6, 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
is_cls_a_cloth Ran
ran on a synthesized input
LeapLabTHU/Pseudo-Q/pseudo_sample_generation/pseudo_query_generation.py
code served (permissive licence) · get_code("ba7829b0ee31bd57")
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 bool, 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
import numpy as np

# Populate the predefined clothing class lists
predefined_hat_cls = ["baseball_cap", "beanie", "top_hat"]
predefined_shirt_cls = ["t_shirt", "polo", "dress_shirt"]
predefined_pants_cls = ["jeans", "khakis", "dress_pants"]
predefined_shoes_cls = ["sneakers", "loafers", "boots"]

# Update the combined cloth list (this mimics the global behavior)
predefined_cloth_cls = predefined_hat_cls + predefined_pants_cls + predefined_shirt_cls + predefined_shoes_cls

# Test with a valid clothing class name
test_cls_name = "jeans"

ARGS = {
    "cls_name": test_cls_name
}
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 bool, 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
is_cls_a_people Ran
ran on a synthesized input
LeapLabTHU/Pseudo-Q/pseudo_sample_generation/pseudo_query_generation.py
code served (permissive licence) · get_code("fd6435078cf70aa8")
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 bool, 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

# Set up the global state that the function depends on
predefined_people_cls = ["person", "human", "man", "woman", "child", "adult"]

# Define arguments for is_cls_a_people function
ARGS = {
    "cls_name": "person"
}
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 bool, 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
read_examples Ran
ran; our contract draft was wrong, not the code
LeapLabTHU/Pseudo-Q/datasets/data_loader.py
pointer only (licence: Apache-2.0); the stored copy get_code serves, from Marloweeee/BARE, is pointer only (licence: NONE) · get_code("e4f50d4eeb6144e9")
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 Marloweeee/BARE, 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.

resize_according_to_long_side Ran
ran; our contract draft was wrong, not the code
LeapLabTHU/Pseudo-Q/datasets/transforms.py
pointer only (licence: Apache-2.0); the stored copy get_code serves, from Marloweeee/BARE, is pointer only (licence: NONE) · get_code("22d10e2294025703")
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 Marloweeee/BARE, 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.

resize_according_to_short_side Ran
ran; our contract draft was wrong, not the code
LeapLabTHU/Pseudo-Q/datasets/transforms.py
pointer only (licence: Apache-2.0); the stored copy get_code serves, from Marloweeee/BARE, is pointer only (licence: NONE) · get_code("553ec6dfb15ff83b")
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 Marloweeee/BARE, 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.

yolo_loss Ran
ran; our contract draft was wrong, not the code
LeapLabTHU/Pseudo-Q/utils/loss_utils.py
pointer only (licence: Apache-2.0); the stored copy get_code serves, from Marloweeee/BARE, is pointer only (licence: NONE) · get_code("105ac2dc1417db62")
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 Marloweeee/BARE, 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.

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 12 distinct functions with an identity, 8 are also present in other papers whose code we hold (3 with at least one paper dated earlier, 0 the same month, 7 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
read_examples 7 other papers
4 dated earlier, 3 later
1908.06354 (2019-08)
A Fast and Accurate One-Stage Approach to Visual Grounding
2008.01059 (2020-08)
Improving One-stage Visual Grounding by Recursive Sub-query Construction
2104.08541 (2021-04)
TransVG: End-to-End Visual Grounding with Transformers
and 4 more not named here
convert_examples_to_features 6 other papers
4 dated earlier, 2 later
1908.06354 (2019-08)
A Fast and Accurate One-Stage Approach to Visual Grounding
2008.01059 (2020-08)
Improving One-stage Visual Grounding by Recursive Sub-query Construction
2104.08541 (2021-04)
TransVG: End-to-End Visual Grounding with Transformers
and 3 more not named here
resize_according_to_long_side 3 other papers
3 later
aaai_28415 (2024 (year only))
p-Laplacian Adaptation for Generative Pre-trained Vision-Language Models
2409.13609 (2024-09)
MaPPER: Multimodal Prior-guided Parameter Efficient Tuning for Referring Expression Comprehension
2601.01526 (2026-01)
BARE: Towards Bias-Aware and Reasoning-Enhanced One-Tower Visual Grounding
resize_according_to_short_side 3 other papers
3 later
aaai_28415 (2024 (year only))
p-Laplacian Adaptation for Generative Pre-trained Vision-Language Models
2409.13609 (2024-09)
MaPPER: Multimodal Prior-guided Parameter Efficient Tuning for Referring Expression Comprehension
2601.01526 (2026-01)
BARE: Towards Bias-Aware and Reasoning-Enhanced One-Tower Visual Grounding
yolo_loss 3 other papers
3 later
2409.04999 (2024-09)
Visual Grounding with Multi-modal Conditional Adaptation
2409.13609 (2024-09)
MaPPER: Multimodal Prior-guided Parameter Efficient Tuning for Referring Expression Comprehension
2601.01526 (2026-01)
BARE: Towards Bias-Aware and Reasoning-Enhanced One-Tower Visual Grounding
build_vl_transformer 2 other papers
2 later
2409.04999 (2024-09)
Visual Grounding with Multi-modal Conditional Adaptation
2409.13609 (2024-09)
MaPPER: Multimodal Prior-guided Parameter Efficient Tuning for Referring Expression Comprehension
crop 2 other papers
2 later
aaai_28415 (2024 (year only))
p-Laplacian Adaptation for Generative Pre-trained Vision-Language Models
2409.13609 (2024-09)
MaPPER: Multimodal Prior-guided Parameter Efficient Tuning for Referring Expression Comprehension
im_list_to_blob 1 other paper
1 dated earlier
2107.12309 (2021-07)
Spatial-Temporal Transformer for Dynamic Scene Graph Generation

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

Visual grounding, i.e., localizing objects in images according to natural language queries, is an important topic in visual language understanding. The most effective approaches for this task are based on deep learning, which generally require expensive manually labeled image-query or patch-query pairs. To eliminate the heavy dependence on human annotations, we present a novel method, named Pseudo-Q, to automatically generate pseudo language queries for supervised training. Our method leverages an off-the-shelf object detector to identify visual objects from unlabeled images, and then language queries for these objects are obtained in an unsupervised fashion with a pseudo-query generation module. Then, we design a task-related query prompt module to specifically tailor generated pseudo language queries for visual grounding tasks. Further, in order to fully capture the contextual relationships between images and language queries, we develop a visual-language model equipped with multi-level cross-modality attention mechanism. Extensive experimental results demonstrate that our method has two notable benefits: (1) it can reduce human annotation costs significantly, e.g., 31% on RefCOCO [65] without degrading original model's performance under the fully supervised setting, and (2) without bells and whistles, it achieves superior or comparable performance compared to state-of-theart weakly-supervised visual grounding methods on all the five datasets we have experimented. Code is available at https://github.com/LeapLabTHU/Pseudo-Q.

For agents

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

get_harvested_code_for_paper("2203.08481")
get_code_for_paper("2203.08481")
have("2203.08481")

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

curl https://syntology.ai/api/ran/2203.08481.json

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

[![Syntology run record](https://syntology.ai/api/ran/2203.08481.svg)](https://syntology.ai/paper/2203.08481)

Connect an agent — have() is free.