SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2303.02153 · ICCV · 2023

Unleashing Text-to-Image Diffusion Models for Visual Perception

Yongming Rao, Wenliang Zhao, Benlin Liu, Jiwen Lu, Jie Zhou, Zuyan Liu

arXiv · PDF · Open in the Atlas

Code that ran

5 of 5 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, 2 violated it and 0 have no contract verdict; 2 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/2303.02153.json · badge: /api/ran/2303.02153.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
wl-zhao/VPD canonical (classifier label)
classifier basis not recorded on the link
5 ran · 0 no recorded run
FunctionStatusWhere it lives
cosine_scheduler Ran
ran, honoured the contract we drafted
wl-zhao/VPD/depth/utils.py
pointer only (licence: MIT); the stored copy get_code serves, from boyazeng/understand_bias, is pointer only (licence: NONE) · get_code("b700fb0a3496b255")
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 boyazeng/understand_bias, 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.

default Ran
ran, violated the contract we drafted
wl-zhao/VPD/vpd/models.py
code served (permissive licence) · get_code("424012cb37b31172")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Sonnet 4.5, prompt python_fixture_v1) 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.

from inspect import isfunction

def exists(val):
    return val is not None

def default(val, d):
    if exists(val):
        return val
    return d() if isfunction(d) else d

# Test case 1: val exists (not None), d is a regular value
ARGS_1 = {
    'val': 42,
    'd': 100
}

# Test case 2: val is None, d is a regular value
ARGS_2 = {
    'val': None,
    'd': 100
}

# Test case 3: val is None, d is a function
def default_func():
    return 200

ARGS_3 = {
    'val': None,
    'd': default_func
}

# Test case 4: val exists, d is a function
ARGS_4 = {
    'val': "existing_value",
    'd': lambda: "default_value"
}

# Use one of the test cases as the final ARGS
ARGS = ARGS_3
Setup block drafted by
a language model, Claude Sonnet 4.5, prompt python_fixture_v1
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
exists Ran
ran, violated the contract we drafted
wl-zhao/VPD/vpd/models.py
pointer only (licence: MIT); the stored copy get_code serves, from ThomasMrY/VCT, is pointer only (licence: NONE) · get_code("aa5486a3650902d8")
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 ThomasMrY/VCT, 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_num_layer_for_swin Ran
ran; our contract draft was wrong, not the code
wl-zhao/VPD/depth/models_depth/optimizer.py
code served (permissive licence) · get_code("6907f72868217c02")
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 torch
import numpy as np

# Analyze the function: it takes var_name (str), num_max_layer (int), and layers_per_stage (list/tuple)
# The function processes variable names and returns layer indices based on Swin transformer architecture

# Test cases to understand the function better:
# 1. var_name with 'encoder' prefix - gets replaced with 'backbone'
# 2. Special token/embedding names return 0
# 3. patch_embed returns 0
# 4. layers with blocks structure - calculates layer_id based on stage and block position
# 5. layers with downsample structure - calculates based on cumulative layers
# 6. other names return num_max_layer - 1

# Create valid test arguments
var_name_test = "encoder.layers.0.blocks.2.norm1.weight"
num_max_layer_test = 25
layers_per_stage_test = [2, 2, 6, 2]  # Typical Swin-T configuration

ARGS = {
    "var_name": var_name_test,
    "num_max_layer": num_max_layer_test,
    "layers_per_stage": layers_per_stage_test
}
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 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
uniq Ran
ran; our contract draft was wrong, not the code
wl-zhao/VPD/vpd/models.py
pointer only (licence: MIT); the stored copy get_code serves, from JingWu321/MUNBa, is pointer only (licence: NONE) · get_code("9a299fe5ae09e407")
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 JingWu321/MUNBa, 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 5 samples lifted from this paper's linked repositories, 5 carry a function identity and 0 do not. Of the 5 distinct functions with an identity, 5 are also present in other papers whose code we hold (5 with at least one paper dated earlier, 3 the same month, 5 later) and 0 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
exists 486 other papers
150 dated earlier, 15 the same month, 315 later, 4 same year, order not determinable, 2 undated · 78 only as an assembly brick
1603.08155 (2016-03)
Perceptual Losses for Real-Time Style Transfer and Super-Resolution
1611.07004 (2016-11)
Image-to-Image Translation with Conditional Adversarial Networks
1804.04637 (2018-04)
EMBER: An Open Dataset for Training Static PE Malware Machine Learning Models
and 483 more not named here
default 207 other papers
38 dated earlier, 6 the same month, 161 later, 2 same year, order not determinable · 41 only as an assembly brick
1711.11585 (2017-11)
High-Resolution Image Synthesis and Semantic Manipulation with Conditional GANs
2006.11239 (2020-06)
Denoising Diffusion Probabilistic Models
2010.08162 (2020-10)
SidechainNet: An All-Atom Protein Structure Dataset for Machine Learning
and 204 more not named here
uniq 115 other papers
9 dated earlier, 5 the same month, 99 later, 2 same year, order not determinable
2112.10752 (2021-12)
High-Resolution Image Synthesis with Latent Diffusion Models
2204.06125 (2022-04)
Hierarchical Text-Conditional Image Generation with CLIP Latents
2206.02779 (2022-06)
Blended Latent Diffusion
and 112 more not named here
cosine_scheduler 26 other papers
7 dated earlier, 17 later, 2 same year, order not determinable
1905.02244 (2019-05)
Searching for MobileNetV3
136810051 (2022 (year only))
AMixer: Adaptive Weight Mixing for Self-Attention Free Vision Transformers
136860514 (2022 (year only))
SLIP: Self-supervision meets Language-Image Pre-training
and 23 more not named here
get_num_layer_for_swin 2 other papers
1 dated earlier, 1 later
2205.13543 (2022-05)
Revealing the Dark Secrets of Masked Image Modeling
2312.14733 (2023-12)
Harnessing Diffusion Models for Visual Perception with Meta Prompts

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

Diffusion models (DMs) have become the new trend of generative models and have demonstrated a powerful ability of conditional synthesis. Among those, text-to-image diffusion models pre-trained on large-scale image-text pairs are highly controllable by customizable prompts. Unlike the unconditional generative models that focus on lowlevel attributes and details, text-to-image diffusion models contain more high-level knowledge thanks to the visionlanguage pre-training. In this paper, we propose VPD (Visual Perception with pre-trained Diffusion models), a new framework that exploits the semantic information of a pretrained text-to-image diffusion model in visual perception tasks. Instead of using the pre-trained denoising autoencoder in a diffusion-based pipeline, we simply use it as a backbone and aim to study how to take full advantage of the learned knowledge. Specifically, we prompt the denoising decoder with proper textual inputs and refine the text features with an adapter, leading to a better alignment to the pre-trained stage and making the visual contents interact with the text prompts. We also propose to utilize the cross-attention maps between the visual features and the text features to provide explicit guidance. Compared with other pre-training methods, we show that vision-language pre-trained diffusion models can be faster adapted to downstream visual perception tasks using the proposed VPD. Extensive experiments on semantic segmentation, referring image segmentation, and depth estimation demonstrate the effectiveness of our method. Notably, VPD attains 0.254 RMSE on NYUv2 depth estimation and 73.3% oIoU on RefCOCO-val referring image segmentation, establishing new records on these two benchmarks. Code is available at https://github.com/wl-zhao/VPD.

For agents

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

get_harvested_code_for_paper("2303.02153")
get_code_for_paper("2303.02153")
have("2303.02153")

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

curl https://syntology.ai/api/ran/2303.02153.json

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

[![Syntology run record](https://syntology.ai/api/ran/2303.02153.svg)](https://syntology.ai/paper/2303.02153)

Connect an agent — have() is free.