Files changed (1) hide show
  1. README.md +10 -7
README.md CHANGED
@@ -4,23 +4,26 @@ license: mit
4
  tags:
5
  - Doc-UFCN
6
  - PyTorch
7
- - Object detection
 
 
 
8
  metrics:
9
  - IoU
10
  - F1
11
  - AP@.5
12
  - AP@.75
13
  - AP@[.5,.95]
 
14
  ---
15
 
 
16
 
17
- # Hugin-Munin line detection
18
-
19
- The Hugin-Munin line detection model predicts text lines from Hugin-Munin document images. This model was developed during the [HUGIN-MUNIN project](https://hugin-munin-project.github.io/).
20
 
21
  ## Model description
22
 
23
- The model has been trained using the Doc-UFCN library on Hugin-Munin document images.
24
  It has been trained on images with their largest dimension equal to 768 pixels, keeping the original aspect ratio.
25
  The model predicts two classes: vertical and horizontal text lines.
26
 
@@ -29,7 +32,7 @@ The model predicts two classes: vertical and horizontal text lines.
29
  The model achieves the following results:
30
 
31
  | set | class | IoU | F1 | AP@[.5] | AP@[.75] | AP@[.5,.95] |
32
- | ----- | ---------- | ----- | ----- | ------- | -------- | ----------- |
33
  | train | vertical | 88.29 | 89.67 | 71.37 | 33.26 | 36.32 |
34
  | | horizontal | 69.81 | 81.35 | 91.73 | 36.62 | 45.67 |
35
  | val | vertical | 73.01 | 75.13 | 46.02 | 4.99 | 15.58 |
@@ -54,4 +57,4 @@ Please refer to the Doc-UFCN library page (https://pypi.org/project/doc-ufcn/) t
54
  pages = {2134-2141},
55
  doi = {10.1109/ICPR48806.2021.9412447}
56
  }
57
- ```
 
4
  tags:
5
  - Doc-UFCN
6
  - PyTorch
7
+ - object-detection
8
+ - dla
9
+ - historical
10
+ - handwritten
11
  metrics:
12
  - IoU
13
  - F1
14
  - AP@.5
15
  - AP@.75
16
  - AP@[.5,.95]
17
+ pipeline_tag: image-segmentation
18
  ---
19
 
20
+ # Doc-UFCN - NorHand v1 - Line detection
21
 
22
+ The NorHand v1 line detection model predicts text lines from NorHand document images. This model was developed during the [HUGIN-MUNIN project](https://hugin-munin-project.github.io/).
 
 
23
 
24
  ## Model description
25
 
26
+ The model has been trained using the Doc-UFCN library on NorHand document images.
27
  It has been trained on images with their largest dimension equal to 768 pixels, keeping the original aspect ratio.
28
  The model predicts two classes: vertical and horizontal text lines.
29
 
 
32
  The model achieves the following results:
33
 
34
  | set | class | IoU | F1 | AP@[.5] | AP@[.75] | AP@[.5,.95] |
35
+ | ----- | ---------- | ----: | ----: | ------: | -------: | ----------: |
36
  | train | vertical | 88.29 | 89.67 | 71.37 | 33.26 | 36.32 |
37
  | | horizontal | 69.81 | 81.35 | 91.73 | 36.62 | 45.67 |
38
  | val | vertical | 73.01 | 75.13 | 46.02 | 4.99 | 15.58 |
 
57
  pages = {2134-2141},
58
  doi = {10.1109/ICPR48806.2021.9412447}
59
  }
60
+ ```