Skip to content

Researcher verdict

Recommended implementation path available

implementation baseline
Benchmark trust: grounded evidence

This page has evidence-backed benchmark findings and a concrete implementation recommendation anchored on tensorflow/models. Use it as an implementation baseline, then validate benchmark parity before adapting it.

Why this page is still worth reading

  • Benchmark findings give you an audit trail for validation before picking an implementation path.
  • A concrete repository path exists via tensorflow/models, so this page can act as a practical starting point.
  • Reproduction risks are surfaced explicitly, which helps decide whether the paper is worth immediate prototyping.

Benchmark trust

Concrete benchmark findings are present and can be audited against the extracted evidence.

Use this page as

Start here when you need the most practical implementation path quickly.

Results & Benchmarks

Freshness tier: cold
Direct + Inferred Evidence
Text recognition from street signs
French Street Name Signs (FSNS)
Accuracy
84.2%
Source: llm grounded
Text recognition from street signs
French Street Name Signs (FSNS)
Accuracy
72.46% (prior SOTA)
Source: llm grounded

Benchmark evidence drill-down

2 findings

Audit each benchmark finding before selecting an implementation path. Evidence refs map to the disclosure section below.

Task Dataset Metric Value Source Evidence refs
Text recognition from street signs French Street Name Signs (FSNS) Accuracy 84.2% llm-grounded
evidencePack.paperSections[id=paper_2]
Text recognition from street signs French Street Name Signs (FSNS) Accuracy 72.46% (prior SOTA) llm-grounded
evidencePack.paperSections[id=paper_2]

Attention-based Extraction of Structured Information from Street View Imagery presents a transformer method.

Use This Implementation Because…

Confidence: high

tensorflow/models is the strongest maintained implementation based on ranking signals. CI workflows are present. License is declared (NOASSERTION).

Open tensorflow/models

Reproduction Risks

  • Dependency manifest is missing

Hardware Notes

Expect multi-day setup/compute for meaningful reproduction based on current guidance.

Evidence disclosure

LLM evidence refs: evidencePack.paperSections[id=paper_2], evidencePack.paperSections[id=paper_4], evidencePack.paperSections[id=paper_12], evidencePack.paperSections[id=paper_18], evidencePack.paperSections[id=paper_11], researcherSummary.implementationRecommendation, repos[0].fullName, evidencePack.repoSources[repoUrl=https://github.com/tensorflow/models].sections[id=docs_1], guidance.riskFlags[0], researcherSummary.reproductionRisks[0], researcherSummary.hardwareNotes[0], researcherSummary.timeToFirstMeaningfulRun, researcherSummary.benchmarkSnapshot[0], paper.title, summary.hasReliableImplementation

Evidence graph: 4 refs, 4 links.

Utility signals: depth 60/100, grounding 85/100, status medium.

Implementation Comparison

Top 3 paths

Compare maintenance quality, reproducibility coverage, and evidence confidence before choosing a reproduction baseline.

tensorflow/models
best maintained
Maintenance: Active
Confidence: High
Reproducibility: Moderate

Official implementation from Papers with Code · Repository link is mentioned in the paper metadata

Stars
77,689
Last push
Mar 6, 2026 (3d ago)
CIReleases

Risk flags

  • No Docker setup
  • Dependency manifest missing
innovationgarage/price-tag-reader
alternative
Maintenance: Stale
Confidence: Low
Reproducibility: Moderate

Matched via arXiv identifier search · Repository appears stale (>24 months since last push)

Stars
13
Last push
Nov 1, 2019 (2320d ago)
Dependencies

Risk flags

  • No push in 12+ months
  • No CI pipeline detected
  • No tagged releases
tanmnguyen/MultiLineAttentionOCR
alternative
Maintenance: Stale
Confidence: Low
Reproducibility: Moderate

Matched via arXiv identifier search · Repository appears stale (>24 months since last push)

Stars
2
Last push
Jan 28, 2024 (771d ago)
Dependencies

Risk flags

  • No push in 12+ months
  • No CI pipeline detected
  • No tagged releases

Paper summary

AI-generated

AI-generated summary grounded in paper metadata and artifact signals.

The paper introduces a neural network that processes Street View images with a CNN, applies a novel attention mechanism to the resulting features, and feeds them into an RNN for structured text extraction. This page includes benchmark evidence for Text recognition from street signs on French Street Name Signs (FSNS). Reproduction guidance focuses on implementation viability and concrete risk controls.

Key contributions

  • The paper introduces a neural network that processes Street View images with a CNN, applies a novel attention mechanism to the resulting features, and feeds them into an RNN for structured text extraction.
  • On the French Street Name Signs (FSNS) dataset, the proposed attention-based CNN-RNN model attains 84.2% accuracy, surpassing the previous state of the art Smith’16 result of 72.46% accuracy.
  • The model architecture supports up to four different views of the same street sign by passing each view through a shared CNN feature extractor and then concatenating the resulting features before attention.
  • The same attention-based CNN-RNN architecture is applied without structural changes to an internal Street View business storefront dataset to extract business name information from single-view images.
  • Quantitative evaluation on the internal Street View business names dataset is not reported because the dataset is not public, and the paper only provides qualitative examples for this setting.

Implementation guidance

Use tensorflow/models first because deterministic ranking and extracted evidence align on implementation viability. Start with the repo setup path, then validate benchmark reproduction before adaptation.

Reproducibility notes

  • Environment setup may fail or produce inconsistent results because the TensorFlow Models repository lacks a complete dependency manifest, requiring manual resolution of library.
  • Training on the full FSNS dataset may be impractical on limited hardware, leading to subpar accuracy compared to the reported 84.2% benchmark.
  • Differences between the current TensorFlow Models API and the original implementation era may cause code incompatibilities or silent behavior changes that affect reproduction.
  • Reproducing experiments on the internal Street View business names dataset is not possible as the dataset is private, limiting validation of the method’s generalization claims.
  • Incorrect preparation or tiling of multi-view FSNS images could break the shared-CNN multi-tile architecture assumption and significantly degrade performance.

Best implementation now

tensorflow/models
Confidence: High
Reproducibility: Moderate

Models and examples built with TensorFlow

Stars: 77,689
Forks: 45,250
Last push: Mar 6, 2026
License: NOASSERTION
Official implementation from Papers with Code
Repository link is mentioned in the paper metadata
Community adoption signal (77689 stars)
License ✓
CI ✓
Deps –
Docker –
  • Selected tensorflow/models as the strongest maintained implementation for new work.
  • Includes CI workflow signals.
  • Repository activity is within the last 24 months.

Reproduction path

Direct

Follow the direct implementation path

  1. 1

    Start with tensorflow/models and validate setup instructions in README.

  2. 2

    Reproduce the baseline result with the provided defaults before modifying hyperparameters.

  3. 3

    Log exact dependency versions and runtime environment for reproducibility.

Time to first repro: a few days
Dependency manifest is missing

Additional implementations

No additional verified repositories beyond the primary recommendation.

These repositories had low-confidence matching signals and are hidden by default.

Hugging Face artifacts

No direct paper-linked artifacts were found. Showing strongest curated related artifacts for faster exploration.

Models

Broaden model search

Transformer attention based extraction structured

Datasets

Broaden dataset search

Transformer dataset attention based extraction structured dataset
Search models Search datasets Search spaces

Research context

Tasks

None detected

Methods

Transformer

Domains

None detected

Evaluation & Human Feedback Data

Open this paper in HFEPX to review benchmark signals, evaluation modes, and human-feedback protocol context.

Open in HFEPX

Explore Similar Papers

Jump to Paper2Code search queries derived from this paper's research context.

Transformer

Need human evaluators for your AI research? Scale annotation with expert AI Trainers.

Post a Job Get a Quote