Recent trends in the development of autonomous vehicles focus on real-time processing of vast amounts of data from various sensors. The data can be acquired using multiple cameras, lidars, ultrasonic sensors, and radars to collect useful information about the state of the traffic and the surroundings. Significant computational power is required to process the data fast enough, and this is even more pronounced in vehicles that not only assist the driver but are capable of fully autonomous driving. This article proposes speed and accuracy improvement of traffic sign detection and recognition in high-definition images, based on focusing on different regions of interest in traffic images. These regions are determined with efficient and parallelized preprocessing of every traffic image, after which convolutional neural network is applied for detection and recognition in parallel on graphics processing units. We employed different “You Only Look Once” (YOLO) architectures as baseline detectors, due to their speed, straightforward architecture, and high accuracy in general object detection tasks. Several preprocessing procedures were proposed, to achieve real-time performance requirement. Our experiments using a large-scale traffic sign dataset show that we can achieve real-time detection in high-definition images with high recognition accuracy.

Detection of small moving objects is an important research area with applications including monitoring of flying insects, studying their foraging behavior, using insect pollinators to monitor flowering and pollination of crops, surveillance of honeybee colonies, and tracking movement of honeybees. However, due to the lack of distinctive shape and textural details on small objects, direct application of modern object detection methods based on convolutional neural networks (CNNs) shows considerably lower performance. In this paper we propose a method for the detection of small moving objects in videos recorded using unmanned aerial vehicles equipped with standard video cameras. The main steps of the proposed method are video stabilization, background estimation and subtraction, frame segmentation using a CNN, and thresholding the segmented frame. However, for training a CNN it is required that a large labeled dataset is available. Manual labelling of small moving objects in videos is very difficult and time consuming, and such labeled datasets do not exist at the moment. To circumvent this problem, we propose training a CNN using synthetic videos generated by adding small blob-like objects to video sequences with real-world backgrounds. The experimental results on detection of flying honeybees show that by using a combination of classical computer vision techniques and CNNs, as well as synthetic training sets, the proposed approach overcomes the problems associated with direct application of CNNs to the given problem and achieves an average F1-score of 0.86 in tests on real-world videos.

Automatic detection of pollen bearing honey bees can provide information important both for pollination monitoring and for assessing health and strength of a beehive. In this paper we analyze some of the methods for detection of pollen bearing honey bees in images obtained at hive entrance. The proposed approach is divided into two parts. In the first part we segment honey bees from images. To this end we analyze two segmentation methods based on color descriptors. Then in the second part we use these segmented regions to classify the bees into two classes with or without pollen. Classification is conducted using SVMs trained on few variations of VLAD-encoded SIFT descriptors. On a dataset of images acquired at the hive entrance, we obtain 0.7971 IoU score for segmentation, and 0.9150 AUC score for classification.

This paper investigates the importance of different parameters of split-brain autoencoder to performance of learned image representations for remote sensing scene classification. We investigate the usage of LAB color space as well as color space created using PCA applied to RGB pixel values. We show that these two spaces give almost equal results, with slight favor towards the LAB color space. We also investigate choices of different quantization methods of color targets and number of quantization bins. We have found that using k-means clustering for quantization works slightly better than using uniform quantization. We also show that even when using really small number of bins it is possible to get only slightly worse results.