Incremental Cross-Modality Deep Learning for Pedestrian Recognition

International audienceIn spite of the large amount of existent methods, pedestrian detection is still an open challenge. In recent years, deep learning classification methods combined with multi-modality images within different fusion schemes achieved the best performance. It was proven that late-fusion scheme out-performs both direct and intermediate integration of modalities for pedestrian recognition. Hence, in this paper, we focus on improving the late-fusion scheme for pedestrian classification on the Daimler stereo vision data set. Each image modality among Intensity, Depth and Flow, is classified by an independent Convolution Neural Network (CNN). The CNN outputs are then fused by a Multi-layer Perceptron (MLP) before the recognition decision. We propose different methods based on Cross-Modality deep learning of CNNs: (1) a correlated model where a unique CNN is learned with Intensity, Depth and respectively Flow images for each frame, (2) an incremental model where a CNN is learned with the first modality images frames, then a second CNN, initialized by transfer learning on the first CNN, is learned on the second modality images frames, and finally a third CNN initialized on the second CNN, is learned on the last modality images frames. The experiments show that the incremental cross-modality deep learning of CNNs allows the improvement of classification performances not only for each independent modality classifier, but also for the multi-modality classifier based on late-fusion. Different learning algorithms were also investigated