OSCR

Vision-Language Models for automated quality control: a benchmarking framework and comprehensive study.

Code ↔ Paper

3 matches between paragraphs of the paper and lines of its authors' code, computed by the harvester (lexical-v1). Click a colored paragraph or line to see its counterpart.

The 3 matches
  1. [1] § Methodology › Evaluation protocol ↔ src/core/response_parser.py, lines 102–125 · score 0.80 · Fuzzy matching, parsing attempts, Parsing Strategy, confidence scores, Regex, parser
  2. [2] § Methodology › Evaluation protocol ↔ src/core/prompt_engine.py, lines 24–172 · score 0.54 · Prompt templates, task instructions, classification
  3. [3] § Experiments › Models and configuration ↔ src/evaluators/gemini.py, lines 65–184 · score 0.51 · top_p, temperature, tokens, max, configurations, models

Paper

Loaded from Europe PMC by your browser, not stored by OSCR: doi.org · Europe PMC

The paper is loaded when this pane is shown.

The authors' code

Python · 516 lines · 21 KB · no license · 1 match

  1. """Universal response parser for handling multiple VLM response formats."""
  3. import json
  4. import re
  5. import logging
  6. from typing import List, Dict, Optional, Any, Tuple
  7. from dataclasses import dataclass
  8. from enum import Enum
  10. logger = logging.getLogger(__name__)
  13. class ParsingStrategy(Enum):
  14. """Available parsing strategies."""
  15. DIRECT_JSON = "direct_json"
  16. CONFIDENCE_SCORES = "confidence_scores"
  17. NESTED_STRUCTURES = "nested_structures"
  18. MIXED_CONTENT = "mixed_content"
  19. REGEX_EXTRACTION = "regex_extraction"
  20. FUZZY_MATCHING = "fuzzy_matching"
  23. @dataclass
  24. class ParseResult:
  25. """Result of parsing a model response."""
  26. success: bool
  27. predicted_class: str
  28. confidence: Optional[float] = None
  29. reasoning: str = ""
  30. strategy_used: Optional[ParsingStrategy] = None
  31. raw_extracted: Optional[Dict] = None
  34. class UniversalResponseParser:
  35. """Handles multiple JSON response formats from different models."""
  37. # Default parsing strategy order (most reliable first)
  38. DEFAULT_STRATEGIES = [
  39. ParsingStrategy.DIRECT_JSON,
  40. ParsingStrategy.CONFIDENCE_SCORES,
  41. ParsingStrategy.NESTED_STRUCTURES,
  42. ParsingStrategy.MIXED_CONTENT,
  43. ParsingStrategy.REGEX_EXTRACTION,
  44. ParsingStrategy.FUZZY_MATCHING
  45. ]
  47. def __init__(self,
  48. allowed_classes: List[str],
  49. strategies: Optional[List[str]] = None,
  50. enable_fuzzy_matching: bool = True,
  51. confidence_threshold: float = 0.0):
  52. """Initialize parser.
  54. Args:
  55. allowed_classes: List of valid class names
  56. strategies: Custom parsing strategy order
  57. enable_fuzzy_matching: Whether to enable fuzzy matching
  58. confidence_threshold: Minimum confidence threshold
  59. """
  60. self.allowed_classes = allowed_classes
  61. self.enable_fuzzy_matching = enable_fuzzy_matching
  62. self.confidence_threshold = confidence_threshold
  64. # Build strategy list
  65. if strategies:
  66. self.strategies = [ParsingStrategy(s) for s in strategies
  67. if s in [e.value for e in ParsingStrategy]]
  68. else:
  69. self.strategies = self.DEFAULT_STRATEGIES.copy()
  71. if not enable_fuzzy_matching:
  72. self.strategies = [s for s in self.strategies if s != ParsingStrategy.FUZZY_MATCHING]
  74. def parse_response(self, response: str) -> ParseResult:
  75. """Try multiple parsing strategies in order of reliability.
  77. Args:
  78. response: Raw model response
  80. Returns:
  81. ParseResult: Parsing result with success status
  82. """
  83. if not response or not response.strip():
  84. return ParseResult(success=False, predicted_class="")
  86. # Try each strategy in order
  87. for strategy in self.strategies:
  88. try:
  89. result = self._apply_strategy(strategy, response)
  90. if result.success:
  91. result.strategy_used = strategy
  92. logger.debug(f"Successfully parsed with strategy: {strategy.value}")
  93. return result
  94. except Exception as e:
  95. logger.debug(f"Strategy {strategy.value} failed: {e}")
  96. continue
  98. # All strategies failed
  99. logger.warning(f"Failed to parse response with any strategy: {response[:100]}...")
  100. return ParseResult(success=False, predicted_class="")
  102. def _apply_strategy(self, strategy: ParsingStrategy, response: str) -> ParseResult:
  103. """Apply a specific parsing strategy.
  105. Args:
  106. strategy: Parsing strategy to use
  107. response: Raw response text
  109. Returns:
  110. ParseResult: Result of parsing attempt
  111. """
  112. if strategy == ParsingStrategy.DIRECT_JSON:
  113. return self._parse_direct_json(response)
  114. elif strategy == ParsingStrategy.CONFIDENCE_SCORES:
  115. return self._parse_confidence_scores(response)
  116. elif strategy == ParsingStrategy.NESTED_STRUCTURES:
  117. return self._parse_nested_structures(response)
  118. elif strategy == ParsingStrategy.MIXED_CONTENT:
  119. return self._parse_mixed_content(response)
  120. elif strategy == ParsingStrategy.REGEX_EXTRACTION:
  121. return self._parse_regex_extraction(response)
  122. elif strategy == ParsingStrategy.FUZZY_MATCHING:
  123. return self._parse_fuzzy_matching(response)
  124. else:
  125. return ParseResult(success=False, predicted_class="")
  127. def _parse_direct_json(self, response: str) -> ParseResult:
  128. """Parse standard JSON classification formats."""
  129. text = self._strip_fences(response)
  131. # Find JSON in response
  132. json_start = text.find('{')
  133. json_end = text.rfind('}') + 1
  135. if json_start < 0:
  136. return ParseResult(success=False, predicted_class="")
  138. # Handle cases where closing brace is missing or not found
  139. if json_end <= json_start:
  140. # Try to extract from opening brace to end of text
  141. json_str = text[json_start:].strip()
  142. # If it doesn't end with }, try adding one
  143. if not json_str.endswith('}'):
  144. json_str = json_str.rstrip() + '}'
  145. else:
  146. json_str = text[json_start:json_end]
  148. # Try parsing the JSON
  149. try:
  150. parsed = json.loads(json_str)
  151. if not isinstance(parsed, dict):
  152. return ParseResult(success=False, predicted_class="")
  154. # Try standard field names
  155. raw_class = (parsed.get('class') or
  156. parsed.get('classification') or
  157. parsed.get('label') or
  158. parsed.get('predicted_label') or
  159. parsed.get('prediction'))
  161. if raw_class:
  162. normalized_class = self._normalize_class_name(raw_class)
  163. if normalized_class:
  164. confidence = parsed.get('confidence', parsed.get('score'))
  165. reasoning = parsed.get('reason', parsed.get('reasoning', ''))
  166. return ParseResult(
  167. success=True,
  168. predicted_class=normalized_class,
  169. confidence=confidence,
  170. reasoning=reasoning,
  171. raw_extracted=parsed
  172. )
  173. except json.JSONDecodeError:
  174. # If simple closing brace addition failed, try more robust repair
  175. try:
  176. # Try to repair truncated JSON by finding the last complete field
  177. repaired_json = self._repair_truncated_json(text[json_start:])
  178. if repaired_json:
  179. parsed = json.loads(repaired_json)
  180. if isinstance(parsed, dict):
  181. raw_class = (parsed.get('class') or
  182. parsed.get('classification') or
  183. parsed.get('label') or
  184. parsed.get('predicted_label') or
  185. parsed.get('prediction'))
  187. if raw_class:
  188. normalized_class = self._normalize_class_name(raw_class)
  189. if normalized_class:
  190. confidence = parsed.get('confidence', parsed.get('score'))
  191. reasoning = parsed.get('reason', parsed.get('reasoning', ''))
  192. return ParseResult(
  193. success=True,
  194. predicted_class=normalized_class,
  195. confidence=confidence,
  196. reasoning=reasoning,
  197. raw_extracted=parsed
  198. )
  199. except json.JSONDecodeError:
  200. pass
  202. return ParseResult(success=False, predicted_class="")
  204. def _parse_confidence_scores(self, response: str) -> ParseResult:
  205. """Parse confidence-based responses like {"Bundle":0.2,"Single-Item":0.8}."""
  206. text = self._strip_fences(response)
  208. # Find JSON in response
  209. json_start = text.find('{')
  210. json_end = text.rfind('}') + 1
  211. if json_start < 0 or json_end <= json_start:
  212. return ParseResult(success=False, predicted_class="")
  214. json_str = text[json_start:json_end]
  216. try:
  217. parsed = json.loads(json_str)
  218. if not isinstance(parsed, dict):
  219. return ParseResult(success=False, predicted_class="")
  221. # Check if this is a confidence-based format (all values are numeric)
  222. if not all(isinstance(v, (int, float)) for v in parsed.values()):
  223. return ParseResult(success=False, predicted_class="")
  225. # Check if all keys are potential class names with numeric values
  226. class_scores = {}
  227. for key, value in parsed.items():
  228. normalized_key = self._normalize_class_name(key)
  229. if normalized_key:
  230. class_scores[normalized_key] = float(value)
  232. if class_scores:
  233. # Find class with highest confidence
  234. best_class = max(class_scores.items(), key=lambda x: x[1])
  235. if best_class[1] >= self.confidence_threshold:
  236. return ParseResult(
  237. success=True,
  238. predicted_class=best_class[0],
  239. confidence=best_class[1],
  240. raw_extracted=parsed
  241. )
  242. else:
  243. # Failed threshold but still parsed correctly
  244. logger.debug(f"Confidence {best_class[1]} below threshold {self.confidence_threshold}")
  245. return ParseResult(success=False, predicted_class="")
  246. except json.JSONDecodeError:
  247. pass
  249. return ParseResult(success=False, predicted_class="")
  251. def _parse_nested_structures(self, response: str) -> ParseResult:
  252. """Parse nested JSON structures."""
  253. text = self._strip_fences(response)
  255. json_start = text.find('{')
  256. json_end = text.rfind('}') + 1
  257. if json_start < 0 or json_end <= json_start:
  258. return ParseResult(success=False, predicted_class="")
  260. json_str = text[json_start:json_end]
  262. try:
  263. parsed = json.loads(json_str)
  264. if not isinstance(parsed, dict):
  265. return ParseResult(success=False, predicted_class="")
  267. # Handle nested classifications array structure
  268. if 'classifications' in parsed:
  269. classifications = parsed['classifications']
  270. if isinstance(classifications, list) and classifications:
  271. first_classification = classifications[0]
  272. if isinstance(first_classification, dict):
  273. raw_class = (first_classification.get('name') or
  274. first_classification.get('class') or
  275. first_classification.get('label'))
  276. if raw_class:
  277. normalized_class = self._normalize_class_name(raw_class)
  278. if normalized_class:
  279. confidence = first_classification.get('confidence',
  280. first_classification.get('score'))
  281. return ParseResult(
  282. success=True,
  283. predicted_class=normalized_class,
  284. confidence=confidence,
  285. raw_extracted=parsed
  286. )
  288. # Handle result wrapper
  289. if 'result' in parsed and isinstance(parsed['result'], dict):
  290. result = parsed['result']
  291. raw_class = (result.get('class') or
  292. result.get('classification') or
  293. result.get('label'))
  294. if raw_class:
  295. normalized_class = self._normalize_class_name(raw_class)
  296. if normalized_class:
  297. return ParseResult(
  298. success=True,
  299. predicted_class=normalized_class,
  300. raw_extracted=parsed
  301. )
  302. except json.JSONDecodeError:
  303. pass
  305. return ParseResult(success=False, predicted_class="")
  307. def _parse_mixed_content(self, response: str) -> ParseResult:
  308. """Parse responses that mix text with JSON."""
  309. # Look for JSON embedded in text
  310. json_patterns = [
  311. r'```json\s*(.*?)\s*```',
  312. r'```\s*(.*?)\s*```',
  313. r'JSON:\s*(\{.*?\})',
  314. r'json\s*(\{.*?\})',
  315. r'(\{[^}]*"(?:class|classification|label)"[^}]*\})'
  316. ]
  318. for pattern in json_patterns:
  319. matches = re.finditer(pattern, response, re.DOTALL | re.IGNORECASE)
  320. for match in matches:
  321. json_candidate = match.group(1).strip()
  322. try:
  323. parsed = json.loads(json_candidate)
  324. if isinstance(parsed, dict):
  325. raw_class = (parsed.get('class') or
  326. parsed.get('classification') or
  327. parsed.get('label'))
  328. if raw_class:
  329. normalized_class = self._normalize_class_name(raw_class)
  330. if normalized_class:
  331. return ParseResult(
  332. success=True,
  333. predicted_class=normalized_class,
  334. raw_extracted=parsed
  335. )
  336. except json.JSONDecodeError:
  337. continue
  339. return ParseResult(success=False, predicted_class="")
  341. def _parse_regex_extraction(self, response: str) -> ParseResult:
  342. """Extract class names using regex patterns."""
  343. # Patterns for finding class names in various contexts
  344. patterns = [
  345. r'"(?:class|classification|label|prediction)"\s*:\s*"([^"]+)"',
  346. r'(?:class|classification|label):\s*"?([^",\s\}]+)"?',
  347. r'classify(?:ing|ied)?\s+(?:as|to)\s+["\']?([^"\'.,\s]+)["\']?',
  348. ]
  350. # First try structured patterns
  351. for pattern in patterns:
  352. matches = re.finditer(pattern, response, re.IGNORECASE)
  353. for match in matches:
  354. candidate = match.group(1).strip()
  355. normalized_class = self._normalize_class_name(candidate)
  356. if normalized_class:
  357. return ParseResult(
  358. success=True,
  359. predicted_class=normalized_class
  360. )
  362. # Only try direct class name matching if it's clearly in a classification context
  363. classification_context_pattern = r'\b(?:is|are|classify|class|label|prediction|result).*?\b(Bundle|Single-Item)\b'
  364. matches = re.finditer(classification_context_pattern, response, re.IGNORECASE)
  365. for match in matches:
  366. candidate = match.group(1).strip()
  367. normalized_class = self._normalize_class_name(candidate)
  368. if normalized_class:
  369. return ParseResult(
  370. success=True,
  371. predicted_class=normalized_class
  372. )
  374. return ParseResult(success=False, predicted_class="")
  376. def _parse_fuzzy_matching(self, response: str) -> ParseResult:
  377. """Last resort fuzzy matching of class names."""
  378. response_lower = response.lower()
  380. # Score each class based on presence in response
  381. class_scores = {}
  382. for class_name in self.allowed_classes:
  383. class_lower = class_name.lower()
  384. score = 0
  386. # Exact match (highest score)
  387. if class_lower in response_lower:
  388. score += 10
  390. # Partial matches for compound words
  391. if '-' in class_lower:
  392. parts = class_lower.split('-')
  393. for part in parts:
  394. if part in response_lower:
  395. score += 3
  397. # Enhanced Single/Bundle specific fuzzy logic
  398. if class_lower == 'single-item' or 'single' in class_lower:
  399. single_keywords = ['single', 'one', 'individual', 'solo', 'alone', 'isolated', 'separate']
  400. for keyword in single_keywords:
  401. if keyword in response_lower:
  402. score += 2
  403. elif class_lower == 'bundle' or 'bundle' in class_lower:
  404. bundle_keywords = ['bundle', 'multiple', 'group', 'several', 'many', 'grouped', 'together', 'collection']
  405. for keyword in bundle_keywords:
  406. if keyword in response_lower:
  407. score += 2
  409. # Generic class name matching
  410. class_words = class_name.lower().replace('-', ' ').split()
  411. for word in class_words:
  412. if len(word) > 2 and word in response_lower: # Avoid matching tiny words
  413. score += 1
  415. if score > 0:
  416. class_scores[class_name] = score
  418. if class_scores:
  419. best_class = max(class_scores.items(), key=lambda x: x[1])
  420. # Lower threshold for fuzzy matching
  421. if best_class[1] >= 1:
  422. return ParseResult(
  423. success=True,
  424. predicted_class=best_class[0],
  425. confidence=min(best_class[1] / 10.0, 1.0) # Normalize to 0-1, cap at 1.0
  426. )
  428. return ParseResult(success=False, predicted_class="")
  430. def _normalize_class_name(self, raw_class: str) -> str:
  431. """Normalize class name to allowed classes."""
  432. if not raw_class:
  433. return ""
  435. raw_lower = raw_class.lower().strip()
  437. # Exact case-insensitive match (exact matching first)
  438. for allowed_class in self.allowed_classes:
  439. if raw_lower == allowed_class.lower():
  440. return allowed_class
  442. # Conservative substring matching for common variations
  443. for allowed_class in self.allowed_classes:
  444. allowed_lower = allowed_class.lower()
  446. # Check for partial word matches in allowed_class
  447. allowed_words = allowed_lower.replace('-', ' ').split()
  448. raw_words = raw_lower.replace('-', ' ').split()
  450. # Check if any raw words completely match any allowed words
  451. if any(raw_word in allowed_words for raw_word in raw_words):
  452. return allowed_class
  454. return ""
  456. def _repair_truncated_json(self, json_text: str) -> str:
  457. """Attempt to repair truncated JSON by finding the last complete field.
  459. Args:
  460. json_text: Potentially truncated JSON text
  462. Returns:
  463. str: Repaired JSON string or empty if repair failed
  464. """
  465. try:
  466. # Remove leading/trailing whitespace
  467. text = json_text.strip()
  469. # Must start with opening brace
  470. if not text.startswith('{'):
  471. return ""
  473. # Find the last complete field by looking for patterns like:
  474. # "field": value,
  475. # "field": value
  476. # "field": [...]
  477. # "field": {...}
  479. # Look for the last complete field ending
  480. patterns = [
  481. r'"[^"]+"\s*:\s*"[^"]*"(?:\s*,)?', # String values
  482. r'"[^"]+"\s*:\s*[\d.]+(?:\s*,)?', # Numeric values
  483. r'"[^"]+"\s*:\s*(?:true|false)(?:\s*,)?', # Boolean values
  484. r'"[^"]+"\s*:\s*\[\s*(?:"[^"]*"(?:\s*,\s*"[^"]*")*\s*)?\](?:\s*,)?', # Arrays
  485. ]
  487. last_match_end = 1 # Start after opening brace
  489. for pattern in patterns:
  490. matches = list(re.finditer(pattern, text, re.IGNORECASE))
  491. for match in matches:
  492. if match.end() > last_match_end:
  493. last_match_end = match.end()
  495. if last_match_end > 1:
  496. # Take content up to last complete field
  497. partial_json = text[:last_match_end]
  499. # Remove trailing comma if present
  500. partial_json = re.sub(r',\s*$', '', partial_json)
  502. # Add closing brace
  503. repaired = partial_json + '}'
  505. # Test if it's valid JSON
  506. json.loads(repaired)
  507. return repaired
  509. except (json.JSONDecodeError, re.error):
  510. pass
  512. return ""
  514. def _strip_fences(self, text: str) -> str:
  515. """Remove markdown code fences."""
  516. return re.sub(r"```(?:json)?\s*([\s\S]*?)```", r"\1", text, flags=re.IGNORECASE).strip()

response_parser.py at commit f3449ff, no license · at the source

Overview

Authors: Mohamed Abdelkader1
  1. Robotics & Internet of Things Lab, Prince Sultan University,Riyadh, Saudi Arabia
Institutions: Prince Sultan University (Saudi Arabia)
Journal: Scientific reports, volume 16, issue 1, article 24047
Dates: received 12 January 2026; accepted 22 May 2026; published online 26 May 2026
Type: Research article · Language: English
License: CC BY-NC-ND
Identifiers: DOI 10.1038/s41598-026-55179-4 · PMID 42191872 · PMCID PMC13439064 · OpenAlex W7162443602
Open access: gold, a free copy (OpenAlex)
Status: code verified
Categories: human (organism), other condition (population), methods / tools (subfield)
Methods: Preprocessing, Statistics, Machine learning
Keywords: Intelligent sensing, AI-based sensing, Vision-Language Models, Quality control, Benchmarking, Real-time object detection, Autonomous systems, Multi-modal AI, Cancer, Computational biology and bioinformatics, Engineering, Mathematics and computing
MeSH: Benchmarking*, Image Processing, Computer-Assisted*, Algorithms, Automation, Detection Algorithms, Humans, Quality Control (* major topic)
Topic: Advanced Neural Network Applications (Computer Vision and Pattern Recognition, Computer Science), according to OpenAlex
Funding: Prince Sultan University
Citations: not cited yet (Europe PMC); 45 references in the paper

Abstract

The abstract is not reproduced here: the paper's license (CC BY-NC-ND) does not allow it. Read it in the paper, at the publisher or on Europe PMC.

Repository

Its files are read in the Code ↔ Paper reader above, with 3 matches between paragraphs and lines of code.

mzahana/vlm-bench

License: none: the authors keep all their rights
State: the link answers, verified on 28 September 2026
Evidence: files inventoried
Commit: f3449ff370c1699fe9a01b6c5312670b0de5ca26, 5 June 2026
Languages: Python (51), Shell (1)
Size: 130 files, 52 scripts
Software Heritage: not archived
Found in: “Code availability”
Holds: README, environment (requirements.txt), tests, documentation
Not found: license file, CITATION.cff, continuous integration
Tools: Pillow (13 files), NumPy (5 files), Matplotlib (2 files), PyTorch (1 file), Hugging Face Transformers (1 file)
Availability: 1 check, the latest on 28 September 2026: the link answers
  • 28 September 2026: the link answers
53 files
At the source: github.com/mzahana/vlm-bench

Code availability statement

The paper has a code availability statement. Its license (CC BY-NC-ND) does not allow reproducing it here; in short, from what the harvester recognized in it:

Read it in the paper: doi.org/10.1038/s41598-026-55179-4.

Tracing map

Proposed by the machine: these links were found in the paper and verified at the source, without human review. The map will receive a Zenodo DOI once one of the paper's authors has validated it with their ORCID.

What the map holds:

  • 1 repository of the authors' code, each at its verified commit, with its license and how the link was found in the paper;
  • 52 scripts, each with its path and the digest of its content;
  • 3 matches between paragraphs of the paper and lines of the code (method lexical-v1);
  • neither the text of the paper nor the code itself.

Its JSON (tracing-map.json) is deposited on Zenodo with its DOI once the map is validated.

Data

Datasets cited

Code and data availability statement

The paper has a code and data availability statement. Its license (CC BY-NC-ND) does not allow reproducing it here; in short, from what the harvester recognized in it:

Read it in the paper: doi.org/10.1038/s41598-026-55179-4.

Versions

The history of this record: each version stored by the harvester or made by a correction of its authors or of the maintainers of its code, and what changed in its facts. The texts of the paper (its abstract, its availability statements) are not part of it; versions that changed only those are not listed.

Version 2, 28 September 2026

  • Funding: added Prince Sultan University

Version 1, 28 September 2026: the first record

Recorded: type, language, journal, volume, issue, pages, dates, 1 author, 12 keywords, 7 MeSH terms, 21 references.

Cite

This paper

Abdelkader, M. (2026). Vision-Language Models for automated quality control: a benchmarking framework and comprehensive study. Scientific reports, 16(1), 24047. https://doi.org/10.1038/s41598-026-55179-4

BibTeX

@article{abdelkader2026vision,
author = {Abdelkader, Mohamed},
title = {{Vision-Language Models for automated quality control: a benchmarking framework and comprehensive study}},
journal = {Scientific reports},
year = {2026},
month = may,
volume = {16},
number = {1},
pages = {24047},
publisher = {Nature Publishing Group},
issn = {2045-2322},
doi = {10.1038/s41598-026-55179-4},
url = {https://doi.org/10.1038/s41598-026-55179-4},
pmid = {42191872},
pmcid = {PMC13439064}
}

RIS

TY - JOUR
AU - Abdelkader, Mohamed
TI - Vision-Language Models for automated quality control: a benchmarking framework and comprehensive study
T2 - Scientific reports
J2 - Sci Rep
PY - 2026
DA - 2026/05/26
VL - 16
IS - 1
SP - 24047
SN - 2045-2322
PB - Nature Publishing Group
DO - 10.1038/s41598-026-55179-4
UR - https://doi.org/10.1038/s41598-026-55179-4
LA - en
ER -

CSL-JSON

{
"id": "10.1038/s41598-026-55179-4",
"type": "article-journal",
"title": "Vision-Language Models for automated quality control: a benchmarking framework and comprehensive study",
"container-title": "Scientific reports",
"author": [
{
"family": "Abdelkader",
"given": "Mohamed"
}
],
"container-title-short": "Sci Rep",
"volume": "16",
"issue": "1",
"page": "24047",
"DOI": "10.1038/s41598-026-55179-4",
"PMID": "42191872",
"PMCID": "PMC13439064",
"ISSN": "2045-2322",
"publisher": "Nature Publishing Group",
"URL": "https://doi.org/10.1038/s41598-026-55179-4",
"language": "en",
"issued": {
"date-parts": [
[
2026,
5,
26
]
]
}
}

The tracing map gets a citation of its own once an author has validated it and it has a DOI.

Similar papers

The papers with a page that share the most with this one: the tools found in their code, their categories, datasets, cited references and authors, the rarest counting most.

[1] doi:10.3389/frai.2026.1849571
NeuroPlast: a learnable activation function evaluated under knowledge distillation for medical image classification.
Journal: Frontiers in artificial intelligence
In common: DOI 10.7910/dvn/dbw86t, methods / tools, other condition, 1 reference
[2] doi:10.1371/journal.pcbi.1014263 [code]
MIRAGE: Robust multi-modal architectures translate fMRI-to-image models from vision to mental imagery.
Journal: PLoS computational biology
In common: Hugging Face Transformers, Pillow, PyTorch, 2 other tools, 1 reference
[3] doi:10.7554/elife.107933 [code]
Modality-agnostic decoding of vision and language from fMRI.
Journal: eLife
In common: Hugging Face Transformers, Pillow, PyTorch, 2 other tools, 1 reference
[4] doi:10.1002/hbm.70469 [code]
VarCoNet: A Variability-Aware Self-Supervised Framework for Functional Connectome Extraction From Resting-State fMRI.
Journal: Human brain mapping
In common: Hugging Face Transformers, Pillow, PyTorch, 2 other tools, methods / tools
[5] doi:10.1038/s41467-026-76837-1 [code]
Drug screen and machine learning predict neuroprotective agents in a preclinical human model of childhood dementia.
Journal: Nature communications
In common: Hugging Face Transformers, Pillow, PyTorch, 2 other tools, other condition
[6] doi:10.1162/imag.a.1299 [code]
A modular semantic-structural pipeline for visual decoding from primate spiking data via selective temporal integration.
Journal: Imaging neuroscience (Cambridge, Mass.)
In common: Hugging Face Transformers, Pillow, PyTorch, 2 other tools, methods / tools
[7] doi:10.1111/joa.70203 [code]
Two-step workflow integrating automatic registration and manual refinement for the accurate alignment of serial histological sections in 3D reconstruction.
Journal: Journal of anatomy
In common: Hugging Face Transformers, Pillow, PyTorch, 2 other tools, methods / tools
[8] doi:10.1038/s41597-026-07248-6 [code]
A large-scale fMRI dataset for vision-language semantic association.
Journal: Scientific data
In common: Pillow, PyTorch, Matplotlib, 1 other tool, methods / tools, 1 reference
[9] doi:10.1038/s42003-026-10957-8 [code]
Brain defence by the extracellular matrix protein Cochlin.
Journal: Communications biology
In common: Hugging Face Transformers, Pillow, PyTorch, 2 other tools
[10] doi:10.1038/s41467-026-76098-y [code]
A single computational objective can produce specialization of streams in visual cortex.
Journal: Nature communications
In common: Hugging Face Transformers, Pillow, PyTorch, 2 other tools

Contribute

The authors of this paper can claim it, correct its record and validate its tracing map, and the maintainers of its code (its owner, or a public member of its organization) correct what it says of their repository; anyone signed in can ask for its removal. Every request goes to OSCR's own machine, which answers it; your account page follows them.

Sign in with ORCID to claim this paper as one of its authors, correct its record or validate its tracing map: when the paper's metadata lists your ORCID iD, you are recognized at once. Maintainers of its code: sign in with GitHub, then claim the repository on your account page.

Request its removal

To ask OSCR to remove this record, the copies of its authors' scripts or its tracing map, use the removal request page: signed in, you say who you are, what to remove and why, then review and confirm the request. Published rules decide every request (how).

Discussion, reproductions, activity

Discussion: questions and error reports about this paper and its code, from signed-in readers and its authors. It opens with sign-in.

Reproductions: reports from readers who ran the authors' code: what they reproduced, with which environment, commit and data. It opens with sign-in.

Activity: what happens around this paper: new versions of its record, its map's validation, discussions and reproductions. It opens with sign-in.