I was trying to implement a text processing application using UIMA. But then I found it very convenient to use uimaFIT rather than UIMA.
Is there anything that UIMA has but uimaFIT does not have? Does uimaFIT add support for natural language processing on video and audio?
uimaFIT is an alternative API for UIMA that builds internally on the original UIMA API. Mainly, the uimaFIT API is less verbose and more intuitive.
In terms of functionality, there is very little that uimaFIT adds that is not possible with plain UIMA (e.g. support for nested external resources).
uimaFIT aims to cover typical use-cases, but it is not a full replacement for the original UIMA API. For advanced functionality, it may be necessary at times to use the original UIMA API, e.g. advanced operations on indexes, access to the low-level CAS API, etc.
Like UIMA, uimaFIT can be used with custom type-systems oriented towards video/audio annotation, but it does not add any specific support for video or audio applications.
Disclosure: I am a member of the Apache UIMA project and look mainly after uimaFIT.