SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2307.09749 · IJCAI · 2023

Towards Robust Scene Text Image Super-resolution via Explicit Location Enhancement

Shu-Tao Xia, Hang Guo, Tao Dai, Guanghao Meng

arXiv · PDF · Open in the Atlas

Code that ran

14 of 16 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, 1 violated it and 0 have no contract verdict; 12 failures of our instrument; 2 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/2307.09749.json · badge: /api/ran/2307.09749.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
csguoh/LEMMA canonical (classifier label)
classifier basis: census
14 ran · 2 no recorded run
FunctionStatusWhere it lives
buf2PIL Ran
ran, honoured the contract we drafted
csguoh/LEMMA/dataset/dataset.py
code served (permissive licence) · get_code("0e7558f55e4382d4")
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 six
from PIL import Image
import io

# Create a simple PIL image and convert it to bytes
test_image = Image.new('RGB', (4, 3), color='red')
img_buffer = io.BytesIO()
test_image.save(img_buffer, format='PNG')
imgbuf = img_buffer.getvalue()

# Create a mock transaction object with a get method
class MockTxn:
    def __init__(self, data):
        self.data = data
    
    def get(self, key):
        return self.data[key]

txn = MockTxn({'test_key': imgbuf})

ARGS = {
    'txn': txn,
    'key': 'test_key',
    'type': 'RGB'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
PIL, six — 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
build_output_control_points Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
csguoh/LEMMA/model/tps_spatial_transformer.py
code served (permissive licence) · get_code("c639f82b23c93b90")
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 2, dtype family float, shape (8, 2) 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)

ARGS = {
    "num_control_points": 8,
    "margins": (0.1, 0.15)
}
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 2, dtype family float, shape (8, 2) — shape (8, 2) 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
central_crop Ran
ran; our contract draft was wrong, not the code
csguoh/LEMMA/dataset/dataset.py
code served (permissive licence) · get_code("5145b55a583f368b")
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.

from PIL import Image

# Create a small test image
im = Image.new('RGB', (100, 80))

ARGS = {
    'im': im
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
PIL — 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
clones Ran
ran; our contract draft was wrong, not the code
csguoh/LEMMA/loss/text_focus_loss.py
pointer only (licence: Apache-2.0); the stored copy get_code serves, from pawn-sxy/DCAMA, is pointer only (licence: NONE) · get_code("84ee72bc513b03f8")
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 pawn-sxy/DCAMA, 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.

compute_partial_repr Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
csguoh/LEMMA/model/tps_spatial_transformer.py
pointer only (licence: Apache-2.0); the stored copy get_code serves, from WarBean/tps_stn_pytorch, is pointer only (licence: NONE) · get_code("e1814689377a5c82")
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 WarBean/tps_stn_pytorch, 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.

conv1x1 Ran
ran; our contract draft was wrong, not the code
csguoh/LEMMA/model/resnet.py
pointer only (licence: Apache-2.0); the stored copy get_code serves, from guanyuezhen/ufcd, is pointer only (licence: NONE) · get_code("6ef392f365cc5703")
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 guanyuezhen/ufcd, 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.

conv3x3 Ran
ran; our contract draft was wrong, not the code
csguoh/LEMMA/model/resnet.py
pointer only (licence: Apache-2.0); the stored copy get_code serves, from lixiaotong97/DSU, is pointer only (licence: NONE) · get_code("fac5364e2f53c6db")
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 lixiaotong97/DSU, 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.

conv3x3_block Ran
ran; our contract draft was wrong, not the code
csguoh/LEMMA/model/stn_head.py
pointer only (licence: Apache-2.0); the stored copy get_code serves, from fudanvi/fudanocr, is pointer only (licence: NONE) · get_code("b59cd9ce9c63bd48")
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 fudanvi/fudanocr, 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.

decoder_layer Ran
ran; our contract draft was wrong, not the code
csguoh/LEMMA/model/attention.py
code served (permissive licence) · get_code("138752fa9826a05c")
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 Sequential. 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 torch.nn as nn

torch.manual_seed(0)

ARGS = {
    'in_c': 3,
    'out_c': 16,
    'k': 3,
    's': 1,
    'p': 1,
    'mode': 'nearest',
    'scale_factor': 2,
    'size': None,
}
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 Sequential — a Sequential output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a Sequential output the recorded ndim 1 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
encoder_layer Ran
ran; our contract draft was wrong, not the code
csguoh/LEMMA/model/attention.py
code served (permissive licence) · get_code("6d0a2c5e1e8c4121")
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 Sequential. 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 torch.nn as nn

torch.manual_seed(0)

ARGS = {
    "in_c": 3,
    "out_c": 64,
    "k": 3,
    "s": 2,
    "p": 1
}
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 Sequential — a Sequential output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a Sequential output the recorded ndim 1 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
grid_sample Ran
ran; our contract draft was wrong, not the code
csguoh/LEMMA/model/tps_spatial_transformer.py
code served (permissive licence) · get_code("47ad53774f24d324")
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 4, dtype family float, shape (2, 4, 3, 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 torch.nn.functional as F

torch.manual_seed(0)

# Create a small input tensor (batch=2, channels=4, height=3, width=3)
input_tensor = torch.randn(2, 4, 3, 3)

# Create a grid tensor for grid_sample
# Grid should have shape (batch, height, width, 2) with values in [-1, 1]
grid = torch.zeros(2, 3, 3, 2)
grid[..., 0] = torch.linspace(-1, 1, 3).view(1, 1, 3)
grid[..., 1] = torch.linspace(-1, 1, 3).view(1, 3, 1)

# Create a canvas tensor (same shape as expected output from grid_sample)
canvas = torch.randn(2, 4, 3, 3)

ARGS = {
    "input": input_tensor,
    "grid": grid,
    "canvas": canvas
}
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 4, dtype family float, shape (2, 4, 3, 3) — shape (2, 4, 3, 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
multi_head_attention_forward Ran
ran; our contract draft was wrong, not the code
csguoh/LEMMA/model/transformer.py
code served (permissive licence) · get_code("0ca1572c9755ffd9")
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. 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 torch.nn.functional as F

torch.manual_seed(0)

# Dimensions
L = 3  # target sequence length
S = 4  # source sequence length
N = 2  # batch size
E = 8  # embedding dimension
num_heads = 2
head_dim = E // num_heads

# Create query, key, value tensors
# Shape: (L, N, E), (S, N, E), (S, N, E)
query = torch.randn(L, N, E)
key = torch.randn(S, N, E)
value = torch.randn(S, N, E)

# Projection weights and biases
# in_proj_weight combines q, k, v projections: shape (3*E, E)
in_proj_weight = torch.randn(3 * E, E)
in_proj_bias = torch.randn(3 * E)

# Output projection
out_proj_weight = torch.randn(E, E)
out_proj_bias = torch.randn(E)

# Bias for key and value
bias_k = None
bias_v = None

# Key padding mask: (N, S)
key_padding_mask = torch.zeros(N, S, dtype=torch.bool)
key_padding_mask[0, -1] = True  # mask last position in first batch

# Attention mask: 2D (L, S)
attn_mask = torch.zeros(L, S)

# Static key and value (not used if None)
static_k = None
static_v = None

# Separate projection weights
q_proj_weight = None
k_proj_weight = None
v_proj_weight = None

ARGS = {
    "query": query,
    "key": key,
    "value": value,
    "embed_dim_to_check": E,
    "num_heads": num_heads,
    "in_proj_weight": in_proj_weight,
    "in_proj_bias": in_proj_bias,
    "bias_k": bias_k,
    "bias_v": bias_v,
    "add_zero_attn": False,
    "dropout_p": 0.0,
    "out_proj_weight": out_proj_weight,
    "out_proj_bias": out_proj_bias,
    "training": True,
    "key_padding_mask": key_padding_mask,
    "need_weights": True,
    "attn_mask": attn_mask,
    "use_separate_proj_weight": False,
    "q_proj_weight": q_proj_weight,
    "k_proj_weight": k_proj_weight,
    "v_proj_weight": v_proj_weight,
    "static_k": static_k,
    "static_v": static_v,
}
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 tuple — a tuple output has no array shape of its own; none is 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
rand_crop Ran
ran; our contract draft was wrong, not the code
csguoh/LEMMA/dataset/dataset.py
code served (permissive licence) · get_code("9a57c2af22605595")
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 MockImage. 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.

from PIL import Image
import random

# Create a mock image object with required methods
class MockImage:
    def __init__(self, width=100, height=80):
        self.size = (width, height)
    
    def crop(self, box):
        # box is (x1, y1, x2, y2)
        # Return a new MockImage representing the cropped region
        return MockImage(
            width=int(box[2] - box[0]),
            height=int(box[3] - box[1])
        )

# Set seed for reproducibility
random.seed(42)

# Create test image
test_image = MockImage(width=100, height=80)

ARGS = {
    "im": test_image
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
PIL — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type MockImage — a MockImage output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a MockImage 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
subsequent_mask Ran
ran, violated the contract we drafted
csguoh/LEMMA/loss/text_focus_loss.py
pointer only (licence: Apache-2.0); the stored copy get_code serves, from ruotianluo/Transformer_Captioning, is pointer only (licence: NONE) · get_code("dd03765c7da9ca51")
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 ruotianluo/Transformer_Captioning, 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.

decoder_layer Not yet run
unverified (harvested, no recorded run)
csguoh/LEMMA/model/Position_aware_module.py
code served (permissive licence) · get_code("fe870f1e0bb99261")
How it ran

no invocation recorded.

encoder_layer Not yet run
unverified (harvested, no recorded run)
csguoh/LEMMA/model/Position_aware_module.py
code served (permissive licence) · get_code("f9c8307f0bce54a3")
How it ran

no invocation recorded.

Also present in other papers' code

Of 16 samples lifted from this paper's linked repositories, 16 carry a function identity and 0 do not. Of the 14 distinct functions with an identity, 11 are also present in other papers whose code we hold (11 with at least one paper dated earlier, 2 the same month, 10 later) and 3 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
conv3x3 686 other papers
489 dated earlier, 8 the same month, 182 later, 7 same year, order not determinable · 31 only as an assembly brick
1406.2661 (2014-06)
Generative Adversarial Networks
1409.0575 (2014-09)
ImageNet Large Scale Visual Recognition Challenge
1412.6572 (2014-12)
Explaining and Harnessing Adversarial Examples
and 683 more not named here
conv1x1 431 other papers
263 dated earlier, 10 the same month, 149 later, 8 same year, order not determinable, 1 undated · 38 only as an assembly brick
1409.1556 (2014-09)
Very Deep Convolutional Networks for Large-Scale Image Recognition
1506.01186 (2015-06)
Cyclical Learning Rates for Training Neural Networks
1507.05717 (2015-07)
An End-to-End Trainable Neural Network for Image-based Sequence Recognition and Its Application to Scene Text Recognition
and 428 more not named here
clones 79 other papers
51 dated earlier, 28 later · 11 only as an assembly brick
1411.4555 (2014-11)
Show and Tell: A Neural Image Caption Generator
1612.00563 (2016-12)
Self-critical Sequence Training for Image Captioning
1706.03762 (2017-06)
Attention Is All You Need
and 76 more not named here
subsequent_mask 21 other papers
15 dated earlier, 6 later
1411.4555 (2014-11)
Show and Tell: A Neural Image Caption Generator
1612.00563 (2016-12)
Self-critical Sequence Training for Image Captioning
1707.07998 (2017-07)
Bottom-Up and Top-Down Attention for Image Captioning and Visual Question Answering
and 18 more not named here
compute_partial_repr 5 other papers
4 dated earlier, 1 later · 1 only as an assembly brick
1509.05329 (2015-09)
Recurrent Spatial Transformer Networks
2203.03373 (2022-03)
Adversarial Texture for Fooling Person Detectors in the Physical World
2203.09388 (2022-03)
A Text Attention Network for Spatial Deformation Robust Scene Text Image Super-resolution
and 2 more not named here
encoder_layer 5 other papers
4 dated earlier, 1 later
2103.06495 (2021-03)
Read Like Humans: Autonomous, Bidirectional and Iterative Language Modeling for Scene Text Recognition
136880442 (2022 (year only))
Multi-modal Text Recognition Networks: Interactive Enhancements between Visual and Semantic Features
aaai_19971 (2022 (year only))
Visual Semantics Allow for Textual Reasoning Better in Scene Text Recognition
and 2 more not named here
decoder_layer 4 other papers
3 dated earlier, 1 later
136880442 (2022 (year only))
Multi-modal Text Recognition Networks: Interactive Enhancements between Visual and Semantic Features
aaai_19971 (2022 (year only))
Visual Semantics Allow for Textual Reasoning Better in Scene Text Recognition
2305.05140 (2023-05)
Linguistic More: Taking a Further Step toward Efficient and Accurate Scene Text Recognition
and 1 more not named here
conv3x3_block 3 other papers
2 dated earlier, 1 later
2112.08171 (2021-12)
Text Gestalt: Stroke-Aware Scene Text Image Super-Resolution
2203.09388 (2022-03)
A Text Attention Network for Spatial Deformation Robust Scene Text Image Super-resolution
build_output_control_points 2 other papers
1 dated earlier, 1 later
2203.09388 (2022-03)
A Text Attention Network for Spatial Deformation Robust Scene Text Image Super-resolution
grid_sample 2 other papers
1 dated earlier, 1 later
2203.09388 (2022-03)
A Text Attention Network for Spatial Deformation Robust Scene Text Image Super-resolution
multi_head_attention_forward 2 other papers
2 dated earlier
2103.06495 (2021-03)
Read Like Humans: Autonomous, Bidirectional and Iterative Language Modeling for Scene Text Recognition
2305.05140 (2023-05)
Linguistic More: Taking a Further Step toward Efficient and Accurate Scene Text Recognition

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

Scene text image super-resolution (STISR), aiming to improve image quality while boosting downstream scene text recognition accuracy, has recently achieved great success. However, most existing methods treat the foreground (character regions) and background (non-character regions) equally in the forward process, and neglect the disturbance from the complex background, thus limiting the performance. To address these issues, in this paper, we propose a novel method LEMMA that explicitly models character regions to produce highlevel text-specific guidance for super-resolution. To model the location of characters effectively, we propose the location enhancement module to extract character region features based on the attention map sequence. Besides, we propose the multimodal alignment module to perform bidirectional visual-semantic alignment to generate high-quality prior guidance, which is then incorporated into the super-resolution branch in an adaptive manner using the proposed adaptive fusion module. Experiments on TextZoom and four scene text recognition benchmarks demonstrate the superiority of our method over other state-of-the-art methods. Code is available at https://github.com/csguoh/LEMMA.

For agents

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

get_harvested_code_for_paper("2307.09749")
get_code_for_paper("2307.09749")
have("2307.09749")

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

curl https://syntology.ai/api/ran/2307.09749.json

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

[![Syntology run record](https://syntology.ai/api/ran/2307.09749.svg)](https://syntology.ai/paper/2307.09749)

Connect an agent — have() is free.