Open-set classification is a problem of handling 'unknown' classes that are not contained in the training dataset, whereas traditional classifiers assume that only known classes appear in the test environment. Existing open-set classifiers rely on deep networks trained in a supervised manner on known classes in the training set; this causes specialization of learned representations to known classes and makes it hard to distinguish unknowns from knowns. In contrast, we train networks for joint classification and reconstruction of input data. This enhances the learned representation so as to preserve information useful for separating unknowns from knowns, as well as to discriminate classes of knowns. Our novel Classification-Reconstruction learning for Open-Set Recognition (CROSR) utilizes latent representations for reconstruction and enables robust unknown detection without harming the known-class classification accuracy. Extensive experiments reveal that the proposed method outperforms existing deep open-set classifiers in multiple standard datasets and is robust to diverse outliers. The code is available in https://nae-lab.

For assistance with grazing cattle management, we propose a cattle detection and counting system based on Convolutional Neural Networks (CNNs) using aerial images taken by an Unmanned Aerial Vehicle (UAV). To improve detection performance, we take advantage of the fact that, with UAV images, the approximate size of the objects can be predicted when the UAV's height from the ground can be assumed to be roughly constant. We resize an image to be fed into the CNN to an optimum resolution determined by the object size and the down-sampling rate of the network, both in training and testing. To avoid repetition of counting in images that have large overlaps to adjacent ones and to obtain the accurate number of cattle in an entire area, we utilize a three-dimensional model reconstructed by the UAV images for merging the detection results of the same target. Experiments show that detection performance is greatly improved when using the optimum input resolution with an F-measure of 0.952, and counting results are close to the ground truths when the movement of cattle is approximately stationary compared to that of the UAV's.

Detecting cars in high-resolution aerial images has attracted particular attention in recent years. However, scene complexity, large illumination change and occlusions make the task very challenging. In this paper, we propose a robust and effective framework for car detection from high-resolution aerial imagery. More specifically, we first incorporate multiple diverse and complementary image descriptors, Histogram of Oriented Gradients (HOG), Local Binary Pattern (LBP) and Opponent Histogram. Subsequently taking computational efficiency and runtime complexity into account, we adopt an interactive bootstrapping approach to collect hard negatives for training an intersection kernel support vector machine (IKSVM). After training, detection is performed by exhaustive search. Finally for post-processing, we employ a greedy procedure for eliminating repetitive detections via non-maximum suppression. Furthermore, contextual information is utilized to refine the detections. Experimental results on Vaihingen dataset have demonstrated that the proposed method can achieve state-of-the-art performance in various real scenes.

No abstract

Abstract. Scene categorization in high-resolution satellite images has attracted much attention in recent years. However, high intra-class variations, illuminations and occlusions make the task very challenging. In this paper, we propose a classification model based on a hierarchical fusion of multiple features. Highlights of our work are threefold: (1) we use four discriminative image features; (2) we employ support vector machine with histogram intersection kernel (HIK-SVM) and L1-regularization logistic regression classifier (L1R-LRC) in different classification stages, respectively. The soft probabilities of different features obtained by the HIK-SVM are discriminatively fused and fed into the L1R-LRC to obtain the final results; (3) we conduct an extensive evaluation of different configurations, including different feature fusion schemes and different kernel functions. Experimental analysis show that our method leads to state-of-the-art classification performance on the satellite scenes.