Graph-based semi-supervised learning with Local Binary Patterns for holistic object categorization

• We propose a graph construction scheme that is based on data self-representation.
• The approach adaptively provides graphs without parameter tuning.
• The proposed scheme uses locality-constrained and sparsity encouraged coding.
• Semi-supervised learning experiments were conducted on indoor and outdoor scenes.
• The proposed method can outperform competing methods.

In this paper, we develop a new efficient graph construction algorithm that is useful for many learning tasks. Unlike the main stream for graph construction, our proposed data self-representativeness approach simultaneously estimates the graph structure and its edge weights through sample coding. Compared with the recent ℓ1ℓ1 graph based on sparse coding, our proposed objective function has an analytical solution (based on self-representativeness of data) and thus is more efficient. This paper has two main contributions. Firstly, we introduce a principled Two Phase Weighted Regularized Least Square graph construction method. Secondly, the obtained data graph is used, in a semi-supervised context, in order to categorize detected objects in outdoor and indoor scenes using Local Binary Patterns as image descriptors. In many previous works, LBP descriptors (histograms) were used as feature vectors for object detection and recognition. However, our work exploits them in order to construct adaptive graphs using a self-representativeness coding. The experiments show that the proposed method can outperform competing methods.