Impala Documentation

virtual svm_problem* Impala::Core::Training::TrainDataSrcFeature::MakeSvmProblem ( ) [inline, virtual]

this function loops over the videos in a set, theoretically the walker already does this.

In practice the walker is cumbersome, so we do it ourselves.

Implements Impala::Core::Training::TrainDataSrc.

Definition at line 74 of file TrainDataSrcFeature.h.

References Impala::Core::Training::SvmProblemBuilder::AddFeatureTable(), Impala::Core::DataFactory::FeatureTableCount(), GetFeatureTableWithCache(), Impala::Core::Training::SvmProblemBuilder::MakeProblem(), Impala::Core::Training::TrainDataSrc::mAnnotation, mDataFactory, mFeatureTableCache, and Impala::Core::Training::TrainDataSrc::mSelection.

00075     {
00076         SvmProblemBuilder pb(mAnnotation, mSelection);
00077         for(int i=0 ; i<mDataFactory->FeatureTableCount() ; ++i)
00078         {
00079             ILOG_DEBUG_NODE("getting feature table...");
00080             Feature::FeatureTable* f = GetFeatureTableWithCache(i);
00081             ILOG_DEBUG_NODE("got feature table");
00082             pb.AddFeatureTable(f);
00083         }
00084         mFeatureTableCache.Empty();
00085         return pb.MakeProblem();
00086     }

Here is the call graph for this function: