Sample and Filter: Nonparametric Scene Parsing via Efficient Filtering
This work addresses computational efficiency and data utilization issues in scene parsing for computer vision, offering an incremental improvement over existing nonparametric techniques.
The paper tackles the problem of computational cost and information loss in nonparametric scene parsing by introducing a sample-and-filter strategy that samples labeled superpixels based on image similarity for a balanced set and uses efficient filtering for label transfer, resulting in improved performance over state-of-the-art methods on two benchmark datasets.
Scene parsing has attracted a lot of attention in computer vision. While parametric models have proven effective for this task, they cannot easily incorporate new training data. By contrast, nonparametric approaches, which bypass any learning phase and directly transfer the labels from the training data to the query images, can readily exploit new labeled samples as they become available. Unfortunately, because of the computational cost of their label transfer procedures, state-of-the-art nonparametric methods typically filter out most training images to only keep a few relevant ones to label the query. As such, these methods throw away many images that still contain valuable information and generally obtain an unbalanced set of labeled samples. In this paper, we introduce a nonparametric approach to scene parsing that follows a sample-and-filter strategy. More specifically, we propose to sample labeled superpixels according to an image similarity score, which allows us to obtain a balanced set of samples. We then formulate label transfer as an efficient filtering procedure, which lets us exploit more labeled samples than existing techniques. Our experiments evidence the benefits of our approach over state-of-the-art nonparametric methods on two benchmark datasets.