Improving cascade of classifiers by sliding window alignment in between

Abstract

Abstract-We improve an object detector based on cascade of classifiers by a local alignment of the sliding window. The detector needs to operate on a relatively sparse grid in order to achieve a real time performance on high-resolution images. The proposed local alignment in the middle of the cascade improves its recognition performance whilst retaining the necessary speed. We show that the moment of the alignment matters and discuss the performance in terms of false negatives and false positives. The proposed method is tested on a car detection problem.

Non-Rigid Object Detection with LocalInterleaved Sequential Alignment (LISA)

“…The gain in speed is obtained by the search space reduction. From the point of view of 2D translation search space reduction (sparser detection grid) has already been shown for a cascade of classifiers in [21] and for SDP in [22]. Here we show that we can effectively reduce a high dimensional search space of non-rigid deformation.…”