We've implemented a user interface to demonstrate our Human-Directed OMR technic. Here we exhibit the results for the three major steps in our Ceres OMR system: staff finding, system indentification and measure recognition. For each section, we show examples from different scores which have various types of errors. From these examples, we can see the efficiency and effectiveness of our proposed method.
Page 6
Before User Input
During User Input
After User Input
Notice that the current step is independent of the previous one. Therefore, the previous labels won't affect the current results.
Page 3
Before User Input
During User Input
After User Input
Page 8
Before User Input During User Input After User Input
Page 6
Before User Input
The First User Input
The Second User Input
After User Input
Page 1
Before User Input
The First User Input
The Second User Input
After User Input
Page 1, Staff 4, Mea 3
Initialize the measure
Switch to Candidate View (arrow: stem, line: beam, cross: slur)
The 1st Recognition
User added a slur candidate
The 2nd Recognition
User added "white space"
The 3rd Recognition
Page 3, Staff 5, Measure 2
Initialize the measure
Switch to Candidate View (arrow: stem, line: beam, cross: slur)
The 1st Recognition
User added "3_beam"
The 2nd Recognition
User added "staccato"
The 3rd Recognition
Page 6, Staff 13, Measure 2
Initialize the measure
Switch to Candidate View (arrow: stem, line: beam, cross: slur)
The 1st Recognition
User added "2_beam"
The 2nd Recognition
User added "2_beam"
The 3rd Recognition
User added "2_beam"
The 4th Recognition
User added "staccato"
The 5th Recognition
User added "staccato"
The 6th Recognition