Expected learning outcomes
Gaining a deeper understanding of principles of object annotation with SALIC (also known as PeriCoDe in connection with PERICLES use cases).
What is the target audience?
- Digital Preservation professionals and students who are interested in learning more about cutting edge technologies in this area, but who are ultimately interested in the application of these tools into their own work.
- Researchers exploring solutions for data management, digital preservation and image annotation
- Teachers/trainers in this field
- Solution providers for organisations in demand of solutions for data / repository management and digital preservation
Level of advancement/ prerequisites
Begin to gain a more advanced understanding of image annotation and approaches to developing training data sets.
- Background to PeriCoDe
- Installation and requirements
- Using PeriCoDe
Time required for completion8 hours (incl. reading list)
More detailed information about SALIC can be found in:
E. Chatzilari, S. Nikolopoulos, Y. Kompatsiaris and J. Kittler, “SALIC: Social Active Learning for Image Classification,” in IEEE Transactions on Multimedia, vol. 18, no. 8, pp. 1488-1503, Aug. 2016. (URL:http://dx.doi.org/10.1109/TMM.2016.2565440)
Maronidis, A. et al. (2016). PERICLES Deliverable D4.3: Content Semantics and Use Context Analysis Techniques. http://pericles-project.eu/uploads/files/PERICLES_WP4_D4_3_Content_Semantics_and_Use_Context_Analysis_Techniques_V1.pdf
Although PeriCoDe addresses a very important problem in automatic image annotation, the collection of annotated training data is only one aspect of the process: the other part is the training of a classifier using these data in order to annotate – or classify – unseen images. This is known as machine learning and is performed in the test implementation of SALIC’s wrapper.m – to learn more about this, the interested reader should explore the following (non exhaustive list of) resources: