Linear SVM Classification of Visual Objects using a Dense Image Representation

Lecture slides:

• Overview
• Codebook construction
• Overview
• Linear SVM Classification of Visual Objects using a Dense Image Representation
• Image representation
• Dimensionality reduction
• Image classification
• Keypoint : Clustering the feature space
• Facing unbalanced densities
• Experiments with the PASCAL dataset (test1 only)
• Bicycles
• Cars
• Motorbikes
• Persons
• cars
• Bikes
• Motorbikes
• People
• Comments on the results
• Conclusions
• Suggestion for next competitions

Author: Frederic Jurie, Institut National de Recherche en Informatique et en Automatique