=Paper=
{{Paper
|id=Vol-2816/paper9
|storemode=property
|title=Location of Simple Graphemes in Mediaeval Manuscripts based on Mask R-CNN
|pdfUrl=https://ceur-ws.org/Vol-2816/paper9.pdf
|volume=Vol-2816
|authors=Simone Marinai, Gabriella Pomaro,Claudia Raffaelli,Francesco Scandiffio
|dblpUrl=https://dblp.org/rec/conf/ircdl/MarinaiPRS21
}}
==Location of Simple Graphemes in Mediaeval Manuscripts based on Mask R-CNN==
https://ceur-ws.org/Vol-2816/paper9.pdf
Location of Simple Graphemes in Mediaeval
Manuscripts based on Mask R-CNN?
Simone Marinai1 , Gabriella Pomaro2 , Claudia Raffaelli1 , and Francesco
Scandiffio1
1
University of Florence, Department of Information Engineering
Via di Santa Marta 3, 50139 Florence, Italy
simone.marinai@unifi.it
{claudia.raffaelli,francesco.scandiffio}@stud.unifi.it
2
Società Internazionale per lo Studio del Medioevo Latino
Via Montebello, 7, 50123 Florence, Italy
gabriella.pomaro@sismelfirenze.it
Abstract. In this paper we describe a system for the location of simple
graphemes in mediaeval manuscripts based on the Mask R-CNN convo-
lutional neural network. This is the first step towards the ambitious goal
of providing palaeographers with a powerful tool with which to speed up
and refine the delicate process of dating and determining the origin of
manuscripts. In order to train the network, a new dataset composed of
49 pages of Latin Middle Ages manuscripts has been built. Experimental
results demonstrate that using the Mask R-CNN network, along with a
proper configuration of parameters, leads to good overall outcomes of
classification.
Keywords: Character recognition · Grapheme classification and loca-
tion · Mask R-CNN · Deep learning · Mediaeval Manuscripts · Paleogra-
phy
1 Introduction
The analysis of manuscripts, in particular their dating and localization, repre-
sents the principal way to reconstruct our history before the invention of movable
type printing. Unfortunately, for the large majority of manuscripts there is no
reliable information about their origin and provenance. As a matter of fact, only
after the late 14th Century we have significant quantities of items with associated
dating information in libraries and archives. For this reason, palaeographers use
a variety of methods to determine the age of a manuscript, but they can usu-
ally only provide an approximate period of time about its origin. Among the
methodologies used by palaeographers we can list the study of the material, the
?
Research supported by Fondazione Cassa di Risparmio di Firenze
Copyright c 2021 for this paper by its authors. Use permitted under Creative Com-
mons License Attribution 4.0 International (CC BY 4.0). This volume is published
and copyrighted by its editors. IRCDL 2021, February 18-19, 2021, Padua, Italy.
ink used, and of course the analysis of the language. In the case of mediaeval
manuscripts, researchers have an additional element from which to draw im-
portant information to carry out their dating task: the analysis of the shape of
graphemes and the features that make them up. Indeed, different ways of writ-
ing the same grapheme have spread among the amanuenses, in a way similar to
what happens to us today with cursive and capital letters. Since these graphic
signs have changed several times and spread slowly over the centuries, they are
a trace of how writing has changed over time. By closely observing the changes
in lettering, palaeographers can provide a basic time frame for when the docu-
ment was written. However, some writing styles lasted for a so long time or were
so widespread that they could not provide any useful information for dating.
For these reasons scholars are interested in examining for each manuscript the
copyist’s graphic choices and also the presence or absence of significant graphic
variants. This process is very long and prone to human errors, such as wrong
reading or missing an occurrence of the searched sign. When manuscripts consist
of a large number of pages i.e. hundreds or thousands of pages, it is very difficult
to completely and carefully inspect them because it would be an extremely time
consuming task. Usually, only a small amount of sample pages is analysed in
order to extract the required information. This kind of approach can easily lead
to incorrect dating results due to the fact that copying a manuscript took years
of time during which even the same amanuensis could change its writing style
several times.
For these reasons, a system capable of extracting information about the pres-
ence of certain palaeographic letter variants within a collection of documents can
be of particular use to palaeographers. In this paper we describe the first version
of a grapheme-detection system based on the Mask R-CNN deep neural network
in order to identify and count the occurrences of a specific subset of graphemes in
manuscripts from the Latin Middle Ages. This work is part of the project "Me-
diaeval manuscripts of Tuscany (XIII-XIV centuries): design and development
of software for dating and determination of origin" financed by SISMEL (So-
cietà Internazionale per lo Studio del Medioevo Latino), DINFO (Dipartimento
di Ingegneria dell’Informazione of the University of Florence), and Fondazione
Cassa di Risparmio Firenze.
The remainder of the paper is structured as follows. In Section 2 we present
related work about character recognition and the Mask R-CNN framework. In
Section 3 we describe the building process of the dataset employed in this project.
In Section 4 we present the approaches used to train the network. Finally, Section
5 contains the obtained results and in Section 6 we draw the conclusions.
2 Related Work
In this section we discuss related work concerning the detection of characters
in manuscript images and we provide a summary of the main object detection
approach considered in our work.
2.1 Character detection in manuscripts
Sheng et al. [1] claim that since automatic document reading does not always
allow to fully understand documents, additional techniques that go beyond the
mere use of OCR systems are needed. In particular, with respect to manuscripts,
locating the geographic origin or identifying the writer may be also relevant tasks.
For this reason the authors have decided to develop a particular set of features
that allows to map the pixels composing the characters in an high-dimensional
space, capturing in this way specific information about the characters. The pro-
posed features can be used separately or jointly and are based on the principle
of Joint Feature Distribution (JFD). The goal is to answer four questions in the
field of palaeography about who produced a certain document, which document,
when and where. The proposed features are divided between Textural based fea-
tures and Grapheme based features. The first ones consider the manuscripts as
textual images, extracting statistical information from the text blocks on the
entire image. They capture the curvature and skew characteristic of different
writing styles and typically do not require line or character segmentation. As for
the grapheme-based features, these allow to capture the statistical distribution
of the single character already segmented starting from documents. These are
based on the principle of the JFD to concatenate spatial information to obtain
a larger structure that is faithful to the traced sign. Among the features that
best allow to date a document, in their work the authors highlight CoHinge and
QuadHinge [2], which fall into the category of textural-based features, as well
as the Junction feature (Junclets) [3] with regard to grapheme-based features.
Wick et al. in [4] propose a method for the automatic transcription of lyrics in
mediaeval music manuscripts. The work is based on the open-source OCR en-
gine Calamari [5]. The predictions are made on previously segmented lines of the
original manuscript page using an available pre-trained model or with a custom
model. In [6], Wahlberg presents a method for line segmentation, along with a
set of features that can be used for text recognition, writer identification and
production dates.
2.2 Mask R-CNN
Artifical Neural Networks ,and in particular deep learning architectures, have
bee widely used in to process historical documents [7]. Among other approaches,
Mask R-CNN is one state of the art model for instance segmentation and object
detection, developed by Facebook AI Research group [8]. Mask R-CNN extends
Faster R-CNN (already used to locate words in early printed documents [9])
making use of an extra mask head and is composed by a standard convolu-
tional network for image classification and, on the top of that, an additional
fully convolutional network for semantic segmentation at pixel level on the pro-
posed regions. The network undergoes through two main stages. The first one
is responsible for generating, on the input image, a set of proposals i.e. regions
where there might be an object. The second one is related to the output pro-
duced by the network and is designed for classifying the proposal suggested at
(a) S Dritta. (b) S Tonda. (c) S Documentaria. (d) D Dritta. (e) D Tonda. (f) K.
(g) Z. (h) Z3. (i) C Cedigliata. (j) Et in legatura. (k) Et tachigrafica. (l) T assibilata.
Fig. 1: Graphemes of interest to be detected.
the previous stage, in order to allow bounding boxes and masks generation. At
the end of the process, the result is a bounding box that encloses the recognised
object and a pixel mask placed on it. The detection branch, that is the branch
for classification and bounding box, runs in parallel with a branch used for pre-
dicting segmentation masks, allowing a decoupling of the two tasks. For more
details refer to [8].
3 Building the dataset
It is important to remark that a program aimed at identifying simple graphemes
should not be designed taking into account one specific period for the manuscript
production or one particular region. Rather, it should focus on the peculiarities of
the graphic material to be examined, in which the simple grapheme has a specific
meaning. Even if the approach we propose is not designed to deal with cursive
writings and would not work for historical periods in which the syntagm is more
important than the paradigm, there are whole centuries whose manuscripts can
be suitably analysed and on which we focus our research: they are manuscripts
between the XIth and the XIVth centuries and also documents from the XVth
century.
The manuscripts used in this study are carefully selected from the large
Codex archive that has been built by SISMEL in the last twenty years by cata-
loguing mediaeval manuscripts from Tuscany in the Codex Project3 . In particu-
lar, the data used are based on the ample collection of scanned works accurately
linked to the codicological descriptions. Taking into account the period of time
previously mentioned, we believe that it is important to identify - and compute
the distribution in the manuscripts - of the following graphemes: three variants
of the letter s and two of the letter d; ligature et; tachygraphic et; k; three vari-
ants of z (including the ç). Given the peculiarities of the dataset, we are also
considering to include the graphic sign ti assibilata: this is a ligature of t and i
3
https://www.sismelfirenze.it/index.php/biblioteca-digitale/codex
Fig. 2: Example of an annotated image viewed from the Interactive LabelMe interface.
that is however an isolated symbol. Currently, the data collection process is still
in progress, so this last grapheme will be introduced in a later version of the
dataset. The set of graphemes of interest is shown in Figure 1.
In order to locate and recognize the graphemes of the subset above, it was
necessary to build an appropriate dataset4 with Latin Middle Ages characters
from the period XI-XIV ineunte. The examples were gathered from manuscripts
of the accessible digital databases, giving preference to those originated from
Tuscany. To achieve good training of the neural network, only documents without
excessive signs of wear such as burns, rubs and tears were selected. However,
it should be noticed that online libraries of ancient documents usually expose
only low-quality images to the public. The difficulty of collecting good quality
pages suitable for our purpose inevitably influenced the quantity of images that
make up the dataset. The dataset consists of 49 pages of which only 11 are
completely labelled. Other documents have already been identified but have not
been validated; we plan to include the future release of the dataset a much larger
number of pages and examples.
The occurrences of the graphemes of interest have been manually annotated
through the Interactive LabelMe program [10, 11] (Figure 2) which outputs a
4
The manuscripts that make up the dataset are listed in the Acknowledgements sec-
tion and are available upon request by sending an email to the authors of this paper.
Table 1: Number of labelled graphemes in each split of the dataset grouped by class.
Class Train Validation Test Total
S dritta 1009 441 533 1983
D tonda 933 329 358 1620
S tonda 367 70 48 485
Et tachigrafica 279 39 26 344
D dritta 192 21 47 260
K 61 27 67 155
Et in legatura 106 21 18 145
C cedigliata 66 17 20 103
Z3 40 25 20 85
S documentaria 44 15 5 64
Z 16 2 2 20
JSON file containing polygonal segmentations of the graphemes. The JSON files
have been converted into the COCO (Common Objects in Context) format for
ease of use with Mask R-CNN. Since characters are usually drawn very close to
each other, some annotations contain not only the grapheme of interest but also
small portions of adjacent signs. This implies that some of the segmentations
have little noise which was nevertheless deemed acceptable.
Some of the images in the dataset have a very high resolution. This has a
positive effect on learning but is also a challenging element for the amount of
GPU memory required for training. After investigating various cutting methods,
we decided to cut each image into four blocks of the same size. Since manuscripts
do not have a predetermined page structure (in some documents the text is a
continuum without separation into columns while in others it surrounds figures
that can be placed anywhere in the page) all images are processed in the same
way. Annotations along the cut lines are discarded.
Since the dataset is to be used for a very specific task, we decided to rely
only on the content of carefully selected manuscripts, avoiding the use of data
augmentation techniques. As a consequence of this choice, considering also that
the Latin language has some graphemes much more frequent than others, the
number of annotated characters is strongly unbalanced in favour of some common
classes and is almost totally non-existent for other rare - but still important -
palaeographic letter variants (see Table 1). For instance, S dritta and D tonda
are the most frequent classes, making up 70% of the dataset. Such an unbalanced
set of data can create some learning issues. It is indeed highly probable that,
with this configuration, the network will learn well the most common classes,
and not so well the rarest ones. Finally, the dataset has been divided into train,
validation and test sets. Since a complete and reliable ground truth is critical to
a proper performance evaluation, the 11 fully annotated pages have been divided
between validation (5) and test (6). The remaining 38 documents were used for
training.
4 Model Identification
In this section we present the experiments conducted to adjust the hyper-parameters
to be used in the network training, discussing the effects produced by their vari-
ations and explaining how they led us towards the final model.
As previously discussed, this work is based on Mask R-CNN which is one
state of the art convolutional network used for object detection and image seg-
mentation. In particular, we selected the Detectron2 implementation developed
by Facebook AI Research Group [12]. To configure the network we used the
fine-tuning paradigm which consists of initialising the weights with a pre-trained
model. This approach is useful when training the network from scratch is made
difficult by limited amount of data available, as pointed out by a variety of scien-
tific publications [13–18]. The chosen model is a ResNet50 with a FPN backbone
trained for 37 epochs on the COCO dataset. After selecting the model, it was
necessary to refine the training parameters, paying particular attention to learn-
ing rate, the number of iterations and the batch size.
In order to identify the optimal learning rate we have carried out 69 indepen-
dent trainings composed of 500 iterations each, assigning to the i-th training the
lri computed as lri = 0.0001 · i. The aim of the experiment was to compute the
loss calculated on the train at the end of the 500 iterations, selecting the lr with
the highest variation towards the minimum value of loss. We observed that from
iteration i = 40 the loss increases, diverging at iteration 69. With low values
of lr (magnitude of 1e-3) we saw a good reduction, but the loss was still high.
We have therefore decided to keep the learning rate of 3.5e-3 which combines an
overall reduction with the minimum global value of loss.
The learning rate schedule and the total number of iterations have been iden-
tified through an experimental trial and error approach. All other configuration
parameters being equal, we evaluated different scheduling policies and max num-
ber of iterations by comparing the Precision, Recall and F 1 measures calculated
on the validation set. Regarding the policy of variation, the best results have
been achieved by training the network with fixed learning rate of 3.5e-3 for a
total of 1150 iterations, obtaining the following values: Precision = 0.869, Recall
= 0.551, F 1 = 0.675. Similar but slightly worse results were obtained by training
with lr fixed at 3.5e-3 for the first 800 iterations, then proceeding with a lr of
5e−4 for other 400 iterations.
Concerning the number of iterations with fixed learning rate, shorter training
provides a precision in the range of ±0.03 from the one of the selected model.
As an opposite case, training for more than 1150 iterations increases precision
by 5 percentage points, but at the same time negatively affects recall, bringing
it down to below 0.20. This trend is confirmed by the comparison of inference
boxes (Figure 3). An excessive number of iterations increases the confidence and
reduces false positives but at the same time makes the model no longer able to
detect graphemes that were previously retrieved. This behaviour is attributable
(a) Test with iteration parameter set to 1150. (b) Test with iteration parameter set to 1250.
Fig. 3: Prediction results on validation produced with different number of training it-
erations. When iterations exceed 1150, box predictions confidence increases but many
instances previously detected detected are no longer recognised.
to the fact that graphemes of the same class are written in slightly different ways
over the dataset, depending on the style of the writer. For instance, some graphic
signs can be traced in a more slanted way, can be larger than others and more
generally present very personal characteristics related to the hand of the writer.
All of this not to mention the fact that some graphemes are more likely to be
drawn close to each other, making it even more difficult to distinguish them.
Keeping all of this in mind, it is not surprising that an excessively high number
of iterations brings the network to overfit on the style of the grapheme used in
the train set, making it difficult to locate the others.
The last hyperparameter to be tuned is the batch size, i.e. the number of
training samples that are analysed by the network before performing a weight
update. The developers of Mask R-CNN adopted an "image-centric training"
with the consequence that the batch size corresponds to the number of images
analysed by the GPUs for each weight update. We choose a batch size of 16,
thus computing 4 documents at a time since each document is divided into four
images.
5 Result and Analysis
The fine-tuning paradigm discussed in the previous section had been actually
used even at an earlier stage of this work, when the available dataset was approx-
imately only 25% of the current size. Considering the dataset expansion work
carried out over the last few months, we questioned the usefulness of initialising
the network with a pre-trained model, thus investigating alternative methods of
initialisation. For this reason, inspired by [19, 20], we have made experiments
Table 2: Number of annotations of the two baseline trainings grouped by class.
Class Initial Train set Current Train set
S dritta 530 1009
D tonda 395 933
S tonda 78 367
Et tachigrafica 85 279
D dritta 28 192
Et in legatura 3 106
C cedigliata 32 66
K 22 61
S documentaria 4 44
Z3 6 40
Z 1 16
Totals 1184 3113
on training from scratch, Furthermore, in this section we analyse how a highly
unbalanced dataset can influence network metrics to appear better than they
actually are.
Table 2 summarises the training examples of the two reference datasets
grouped by classes. It is easy to observe that due to the intrinsic rarity of some
graphemes, the example instances are not properly balanced. Moreover, more
than half of the classes in the initial dataset have fewer than 35 annotations, an
insignificant number compared to the great variety of sign executions that can
be found even within a single manuscript.
5.1 Models comparison
Throughout the analysis of the results we decided to prefer a higher recall even
at the expense of precision, provided that the value of the latter was at least
80%. The reason for this choice is related to the final objective of the research
project. If the automatic grapheme identification and localisation system had a
high recall and low precision it would provide many wrong results, leading the
user to not trust the system and requiring to manually check a large amount of
retrieved data. This would surely discourage the use of the software and therefore
must be avoided. Even the opposite situation - high precision, low recall - would
be counterproductive because it would provide data that is unrepresentative and
not able to satisfy the research, leading the palaeographer to search manually
for important but undetected graphemes. However, we would like to point out
that usually the analysis of the writing is manually carried out and that due to
the complexity and heaviness of the task it is usually done only on a very limited
selection of pages. For this reason obtaining even only a third of the instances
of a manuscript would be a considerable improvement.
Table 3: Comparison of the key metrics calculated on the validation set and test set
on different network configurations.
Validation Test
Precision Recall F1 Precision Recall F1
Network 0H 0.857 0.452 0.592 0.841 0.446 0.583
Network 1H 0.869 0.551 0.675 0.869 0.490 0.624
Network 0L 0.903 0.510 0.651 0.854 0.528 0.653
Network 1L 0.861 0.564 0.682 0.839 0.610 0.706
Considering the two datasets and the two techniques for initialising the
weights, we can identify the following scenarios to which we associate codes
for the sake of brevity: training from scratch on the initial dataset (0L); training
from scratch on the updated dataset (0H); pre-trained weights and small dataset
(1L); pre-trained weights and updated dataset (1H). By applying the approach
discussed in Section 4 we obtained the following 4 models:
– 0L is trained for 1400 iterations with fixed learning rate at 0.0035
– 1L is trained for 1400 iterations with fixed learning rate at 0.0035
– 0H is trained for 800 iterations with fixed learning rate at 0.0035
– 1H is trained for 1150 iterations with fixed learning rate at 0.0035
In the first analysis we make pairwise comparisons between the models group-
ing by weight initialisation method and structure of the training set (Table
3). Comparing 0H and 1H it is evident that the network initialized with pre-
calculated weights has better recall and precision values than its untrained coun-
terpart. This is justified by the fact that although the dataset is better supplied
with examples, these are not sufficient to allow a good training of the network
without the support of a basic model.
Comparing the models 0L and 1L on the validation set it emerges that, given
an equal length of training, initialising with pre-trained weights brings a benefit
in terms of recall (+0.054) at the expense of precision which instead decreases
by 0.042 points. From the analysis of the evaluation metrics, as the number
of iterations varies 0L shows a trend that grows smoothly on both precision
and recall, going into overfitting after iteration 1400. The 1L metrics, on the
other hand, are more abrupt, oscillating several times before reaching the values
previously reported. These behaviours are in line with what was discussed in
Section 4. The trend of the validation set is confirmed by the results obtained
on the test set.
When comparing the results on the test of all the networks, 1L and 0L models
obtain the best scores, ranking first and second respectively for F1 measure. From
these values it may seem that the models trained on the intial dataset are better
than those on the updated version. This would mean that the update of the
dataset made things worse, an unlikely behaviour if we compare the composition
of the two datasets. Although the problem of imbalance is still present, albeit in a
slightly reduced form: more than half of the classes exceeded the 100-annotation
Table 4: Comparison of the precision and recall metrics obtained on the test set and
grouped by classes for both network configurations 1L and 1H.
Network 1L Network 1H
Class Precision Recall Precision Recall Test set
S dritta 0.809 0.835 0.869 0.650 533
D tonda 0.919 0.664 0.952 0.332 358
S tonda 1.0 0.042 0.823 0.292 67
Et tachigrafica 0.611 0.423 0.733 0.423 48
D dritta 0.0 0.0 1.0 0.149 47
K 0.0 0.0 0.704 0.567 26
Et in legatura 0.0 0.0 0.882 0.833 20
C cedigliata 0.0 0.0 0.857 0.3 20
Z3 0.0 0.0 0.0 0.0 18
S documentaria 0.0 0.0 0.0 0.0 5
Z 0.0 0.0 0.0 0.0 2
threshold, becoming more significant in the training phase. Since this can only
be a positive fact, a deeper analysis is necessary.
First of all, we note that in this context it is much more useful and accurate to
assess precision and recall separately for each class, as in Table 4. This method of
analysis is necessary to take into account the different probabilities of occurrence
of Latin graphemes, element that inevitably affects the number of examples in
the dataset. However, every grapheme in the set of interest has relevance and
this is why it would be unacceptable to produce good results on only a subset
of the selected classes.
From the results grouped by class (Table 4) it is clear that 1L cannot provide
information on more than half of the characters and is therefore an unsatisfac-
tory model. This result can be explained by looking at the structure of the initial
training set: the 1L model has been able to specialise exclusively on the recog-
nition of the first four characters with the largest number of examples.
Comparing 1H and 1L we can say that having a lower recall on the most common
classes and a higher recall for all the others is a good indicator that the dataset
update has succeeded in preventing the network from specialising on a subset
of characters, bringing us closer to the final goal. Certainly further work needs
to be done to increase the number of examples relating to the graphemes Z3,
S documentary, Z, which are currently not recognised by the network because
they are last in terms of number of annotations.
Summarizing, the information on the performance of the models contained
in Table 3 expresses a value that may seem absolute but that in reality is closely
related to the structure of the training set used. It is therefore incorrect to use
the values in Table 3 to compare L models with H models and say that L models
are preferable to H models because they achieve better metrics. Instead, it is
correct to say that 1L and 1H are better than 0L and 0H respectively, i.e. that
initialization with pretrained weights produced better results than initialization
from scratch.
6 Conclusions
The content of this paper is part of a larger project which aims to provide
palaeographers with a software able to classify and locate simple graphemes
within mediaeval manuscripts. The major benefit of an automatic detection will
be to replace the time-consuming process of manual analysis with a quick and
easy way of obtaining information about simple graphemes contained within
entire document collections. The core of the system is a Mask R-CNN network
trained to recognise a specific subset of graphemes. The training phase was
carried out on a new dataset built by manually labelling images of manuscript
from the period XI to XIV. The results discussed in Section 5 show that among
the proposed models the best results are achieved by the pre-trained network.
This is justified by the fact that the amount of data available needs to be further
increased and better balanced before dropping the use of pre-trained weights.
Training from scratch provided satisfactory results, although slightly worse than
its pre-trained counterpart.
The first future development to be carried out concerns the improvement of
the dataset. Expanding the dataset by adding examples for rarer classes would
help in increasing the recall for those classes. In order to enhance the quality of
the training dataset another technique that could be applied is the one of data
augmentation in favour of those classes with fewer examples.
Acknowledgements
This work is partially supported by the Fondazione Cassa di Risparmio di
Firenze that funded the project I manoscritti medievali della Toscana (sec. XIII-
XIV): progettazione e sviluppo di un software per la datazione e la determina-
zione di origine granted to SISMEL.
We would like to thank the following libraries for providing us the documents:
– Barcellona, Biblioteca de Catalunya: 639 f. non precisabile
– Bologna, Biblioteca Universitaria: 1746 ff. 7, 8, 10, 30, 42, 52
– Berlin, Staatsbibliothek zu Berlin: Rehdiger 227 f. 10r; Phillips 1716 12v, f.
39r
– Cologny, Fondation Martin: Bodmer 30 f. 12v
– Firenze, Biblioteca della Fondazione E. Franceschini: ms. 2 f. 222v, ignota
– Firenze, Biblioteca Medicea Laurenziana: Plut. 42.23 f. 1r - Plut. 19 dex. 1
f. 7v - Plut. 19 dex. 5 ff. 5v, 6v, 14r, 23v, 55r, 55v, 135v - Plut. 19 dex. 8 f.
5v - Plut. 19 dex. 7 ff. 21v, 89v, 108v - Plut. 30 sin. 3 ff. 42v, 110v, 235v -
Conv.Soppr. 321 ff. 145r, 146r, 150v, 151r - Strozzi 146, f. 2r, 12r
– Firenze, Biblioteca Riccardiana: 222 f. 152r - 269 f. 1r - 323 f. 105v - 327 f.
8r - 1422 f. 70r – 829 f. 12r - 1471 f. 43v
– Firenze, Biblioteca Nazionale Centrale: I.III 272-273 f. 32r - C.S D.7.1158 f.
10r - Magl. XII.4, f.17r
– Milano, Biblioteca Ambrosiana: M 76 sup. f. 274
– Pisa, Archivio di Stato: Div. A n. 2 f. 101v
– Siena, Biblioteca Comunale degli Intronati: F.III.3 ff. 2r, 137r.
References
[1] He Sheng and Lambert Schomaker. “Beyond OCR: Multi-faceted under-
standing of handwritten document characteristics”. In: Pattern Recognition
63 (Mar. 2017), pp. 321–333. doi: 10.1016/j.patcog.2016.09.017.
[2] S. He and L. Schomaker. “Co-occurrence Features for Writer Identifica-
tion”. In: 2016 15th International Conference on Frontiers in Handwriting
Recognition (ICFHR). 2016, pp. 78–83. doi: 10.1109/ICFHR.2016.0027.
[3] Lambert Schomaker, Marco Wiering, and He Sheng. “Junction detection
in handwritten documents and its application to writer identification”. In:
Pattern Recognition 48 (June 2015). doi: 10.1016/j.patcog.2015.05.
022.
[4] C. Wick, A. Hartelt, and F. Puppe. “Lyrics Recognition and Syllable As-
signment of Medieval Music Manuscripts”. In: 2020 17th International
Conference on Frontiers in Handwriting Recognition (ICFHR). 2020, pp. 187–
192. doi: 10.1109/ICFHR2020.2020.00043.
[5] Christoph Wick, Christian Reul, and Frank Puppe. Calamari - A High-
Performance Tensorflow-based Deep Learning Package for Optical Char-
acter Recognition. 2018. arXiv: 1807.02004 [cs.CV].
[6] Fredrik Wahlberg. “Interpreting the Script: Image Analysis and Machine
Learning for Quantitative Studies of Pre-modern Manuscripts”. PhD the-
sis. Acta Universitatis Upsaliensis, 2017.
[7] Francesco Lombardi and Simone Marinai. “Deep Learning for Historical
Document Analysis and Recognition - A Survey”. In: J. Imaging 6.10
(2020), p. 110. url: https://doi.org/10.3390/jimaging6100110.
[8] K. He et al. “Mask R-CNN”. In: 2017 IEEE International Conference on
Computer Vision (ICCV). 2017, pp. 2980–2988. doi: 10 . 1109 / ICCV .
2017.322.
[9] Zahra Ziran et al. “Text alignment in early printed books combining deep
learning and dynamic programming”. In: Pattern Recognit. Lett. 133 (2020),
pp. 109–115.
[10] Samuele Capobianco. “Deep Learning Methods for Document Image Un-
derstanding”. PhD thesis. University of Florence, 2020.
[11] Kentaro Wada. labelme: Image Polygonal Annotation with Python. https:
//github.com/wkentaro/labelme. 2016.
[12] Yuxin Wu et al. Detectron2. https://github.com/facebookresearch/
detectron2. 2019.
[13] C. Käding et al. “Fine-Tuning Deep Neural Networks in Continuous Learn-
ing Scenarios”. In: ACCV Workshops. 2016.
[14] Pulkit Agrawal, Ross Girshick, and Jitendra Malik. Analyzing the Perfor-
mance of Multilayer Neural Networks for Object Recognition. 2014. arXiv:
1407.1610 [cs.CV].
[15] R. Girshick et al. “Rich Feature Hierarchies for Accurate Object Detection
and Semantic Segmentation”. In: 2014 IEEE Conference on Computer Vi-
sion and Pattern Recognition. 2014, pp. 580–587. doi: 10 . 1109 / CVPR .
2014.81.
[16] M. Oquab et al. “Learning and Transferring Mid-level Image Representa-
tions Using Convolutional Neural Networks”. In: 2014 IEEE Conference
on Computer Vision and Pattern Recognition. 2014, pp. 1717–1724. doi:
10.1109/CVPR.2014.222.
[17] Steve Branson et al. Bird Species Categorization Using Pose Normalized
Deep Convolutional Nets. 2014. arXiv: 1406.2952 [cs.CV].
[18] Artem Babenko et al. Neural Codes for Image Retrieval. 2014. arXiv:
1404.1777 [cs.CV].
[19] Yuxin Wu et al. Detectron2 Model Zoo and Baselines. 2020. url: https:
//github.com/facebookresearch/detectron2/blob/master/MODEL_
ZOO.md.
[20] Kaiming He, Ross Girshick, and Piotr Dollár. Rethinking ImageNet Pre-
training. 2018. arXiv: 1811.08883 [cs.CV].