-
Notifications
You must be signed in to change notification settings - Fork 7
/
Copy pathRELEASE_NOTES.txt
73 lines (57 loc) · 2.23 KB
/
RELEASE_NOTES.txt
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
-----------------------------
RELEASE NOTES - TranskribusDU
-----------------------------
--- Vulnéraire - 2020-08-07
- added support to evaluate the segmentation (see --eval_.. options)
- added support for oracle evaluation (--edge_oracle) indicating best achievable quality
- added support to record experiences in MLFLOW (see in options)
- fix for constructing graph when objects overlap each other (--g1o option)
- related to --g1o, better support for constructing object's bounding-box
- few bug and upgrade fix (e.g. in TestReport class, due to scipy evolution, ...)
--- Chrysanthème - 2019-11-21
- ICDAR19 papers are reproducible
- major code reorganisation
- Multipage XML bug fixes
- standard projection profile method
- convex hull for cluster Coords
- ECN ensemble bug fix
- various bug fixes
- --server mode
- segmentation task using agglomerative clustering
- Json input
- pipe example
- table reconstruction
- generic features (when no page info)
- edge features reworked
- cluster evaluation metrics
--- Iris - 2019-04-25
- CRF, ECN GAT supported
- conjugate mode supported
- --vld option to specify a validation set, or a ratio of validation graphs
taken from the training set. The best model on validation set is kept.
- --graph option to store the edges in the output XML
- --max_iter applies to all learning methods
- --seed to seed the randomizer with a constant
- dynamic load of the learners
- major code re-organization
- for example of use, see in tasks: DU_TABLE_BIO.py or DU_Table_Row_Edge.py
--- Jonquille - 2017-04-28
- multi-type classification supported
- task ABP Table
--- Héllebore - 2017-04-20
- logit extractor to deal with textual features
- train/test/run options
- warm start
- plot of training loss curve
- show model information
- baseline method included
- cross-validation options
- support for hyperparameter tuning
- task GTBooks
--- Initial Version - 2016-12-13
- task StAZH
---------------------------------------------------------------------------------
Copyright (C) 2016-2019 H. Déjean, JL Meunier
Developed for the EU project READ. The READ project has received funding
from the European Union's Horizon 2020 research and innovation programme
under grant agreement No 674943.