3. Manual Layout Recognition

If some text regions and lines have not been perfectly detected automatically, you can quickly correct them manually

Sometimes it may be necessary to make some manual corrections to automate layout recognition, especially when the layout of your documents is complex.

When manually transcribing or correcting the automatic transcription, you may notice some empty regions, some lines too short, or the automatic recognition has detected two lines instead of one. 

To correct the layout, open the page. The Transkribus editor shows the image and the text, on the right or below the image, according to your view preferences. On the left side of the image, you will find the buttons to add a new line, region and table.

Here all the possible layout options are explained (you can read about them in Transkribus by clicking on the Info button):

• To add new lines, press the line button, click on the image multiple times and finish the line with a double click or enter
• To add new regions, press the region button and click on the image twice to create the area
• New points can be added by clicking in the middle of a line and dragging the new point to a new position
• Press CANC to delete a selected shape
• Hold SHIFT to move and scale shapes
• Press CTRL+L to select all lines of a selected region
• Hold CTRL or Command to select multiple shapes by clicking or by spanning a selection rectangle
• Press TAB to select the next shape (following the reading order)
• Hold A to add points to the beginning or end of a selected line
• Hold V to split a shape vertically
• Hold H to split a shape horizontally
• Hold C to split a shape with a custom line (use arrow keys to change the angle)
• Press M to merge two selected lines
• Press ESC to leave the current mode and enter selection mode
• Press CTRL+C to copy all selected shapes
• Press CTRL+V to paste copied shapes (to the same page or a new one)

It is also possible to edit a shape using the context menu that appears when you right-click on a shape.

Transkribus eXpert (deprecated)

Sometimes it may be necessary to make some manual corrections to automate layout recognition, especially when the layout of your documents is complex.

When manually transcribing or correcting the automatic transcription, you may notice some empty regions, some lines too short, or the automatic recognition has detected two lines instead of one. 

To correct the layout, use the Canvas Menu (to the left of the image). Most of the common edits are:

• Modify the shape of a text region: click inside the region so that it is highlighted, and drag the border of the text region as needed.

• Split a shape into two: you can use the Split Shapes buttons in the Canvas Menu. The “H-button” splits the shape horizontally. The “V” button splits it vertically. The “L-button” allows you to split it with a customisable line.

• Delete a shape: click on the shape you wish to delete, and click the red “Remove a shape” button in the Canvas Menu.

• Merge two shapes: sometimes, the automatic layout analysis creates two shapes where only one is needed. In this case, you can easily merge the two together. Hold down the “CTRL” button on your keyboard and select both shapes (they need to be the same type of shape). Click the “Merges the selected shapes” button in the Canvas Menu.

• Add a text region: click the “+TR” button in the Canvas Menu. Click once to start drawing your region and double-click to finish it.

• Add a line: click the “+L” button in the Canvas Menu. Click once to start drawing your line and double-click to finish it.

• Add a baseline: click the “+BL” button in the Canvas Menu. Click once to start drawing your baseline and double-click to finish it.

• Correct baselines: click on one point of the baseline and drag it to the new position. Remember that baselines are most important for HTR and it is important that they run at the bottom of the words; lines do not need to be corrected.