RESUMEN
Rare-class objects in natural scene images that are usually small and less frequent often convey more important information for scene understanding than the common ones. However, they are often overlooked in scene labeling studies due to two main reasons, low occurrence frequency and limited spatial coverage. Many methods have been proposed to enhance overall semantic labeling performance, but only a few consider rare-class objects. In this work, we present a deep semantic labeling framework with special consideration of rare classes via three techniques. First, a novel dual-resolution coarse-to-fine superpixel representation is developed, where fine and coarse superpixels are applied to rare classes and background areas respectively. This unique dual representation allows seamless incorporation of shape features into integrated global and local convolutional neural network (CNN) models. Second, shape information is directly involved during the CNN feature learning for both frequent and rare classes from the re-balanced training data, and also explicitly involved in data inference. Third, the proposed framework incorporates both shape information and the CNN architecture into semantic labeling through a fusion of probabilistic multi-class likelihood. Experimental results demonstrate competitive semantic labeling performance on two standard datasets both qualitatively and quantitatively, especially for rare-class objects.
RESUMEN
We propose new techniques for joint recognition, segmentation and pose estimation of infrared (IR) targets. The problem is formulated in a probabilistic level set framework where a shape constrained generative model is used to provide a multi-class and multi-view shape prior and where the shape model involves a couplet of view and identity manifolds (CVIM). A level set energy function is then iteratively optimized under the shape constraints provided by the CVIM. Since both the view and identity variables are expressed explicitly in the objective function, this approach naturally accomplishes recognition, segmentation and pose estimation as joint products of the optimization process. For realistic target chips, we solve the resulting multi-modal optimization problem by adopting a particle swarm optimization (PSO) algorithm and then improve the computational efficiency by implementing a gradient-boosted PSO (GB-PSO). Evaluation was performed using the Military Sensing Information Analysis Center (SENSIAC) ATR database, and experimental results show that both of the PSO algorithms reduce the cost of shape matching during CVIM-based shape inference. Particularly, GB-PSO outperforms other recent ATR algorithms, which require intensive shape matching, either explicitly (with pre-segmentation) or implicitly (without pre-segmentation).
RESUMEN
We propose a new integrated target tracking, recognition and segmentation algorithm, called ATR-Seg, for infrared imagery. ATR-Seg is formulated in a probabilistic shape-aware level set framework that incorporates a joint view-identity manifold (JVIM) for target shape modeling. As a shape generative model, JVIM features a unified manifold structure in the latent space that is embedded with one view-independent identity manifold and infinite identity-dependent view manifolds. In the ATR-Seg algorithm, the ATR problem formulated as a sequential level-set optimization process over the latent space of JVIM, so that tracking and recognition can be jointly optimized via implicit shape matching where target segmentation is achieved as a by-product without any pre-processing or feature extraction. Experimental results on the recently released SENSIAC ATR database demonstrate the advantages and effectiveness of ATR-Seg over two recent ATR algorithms that involve explicit shape matching.