Estimating Efforts and Success of Symmetry-Seeing Machines by Use of Synthetic Data

Representative input data are a necessary requirement for the assessment of machine-vision systems. For symmetry-seeing machines in particular, such imagery should provide symmetries as well as asymmetric clutter. Moreover, there must be reliable ground truth with the data. It should be possible to estimate the recognition performance and the computational efforts by providing different grades of difficulty and complexity. Recent competitions used real imagery labeled by human subjects with appropriate ground truth. The paper at hand proposes to use synthetic data instead. Such data contain symmetry, clutter, and nothing else. This is preferable because interference with other perceptive capabilities, such as object recognition, or prior knowledge, can be avoided. The data are given sparsely, i.e., as sets of primitive objects. However, images can be generated from them, so that the same data can also be fed into machines requiring dense input, such as multilayered perceptrons. Sparse representations are preferred, because the author’s own system requires such data, and in this way, any influence of the primitive extraction method is excluded. The presented format allows hierarchies of symmetries. This is important because hierarchy constitutes a natural and dominant part in symmetry-seeing. The paper reports some experiments using the author’s Gestalt algebra system as symmetry-seeing machine. Additionally included is a comparative test run with the state-of-the-art symmetry-seeing deep learning convolutional perceptron of the PSU. The computational efforts and recognition performance are assessed.

Mining Hard Negative Samples for SAR-Optical Image Matching Using Generative Adversarial Networks

In this paper, we propose a generative framework to produce similar yet novel samples for a specified image. We then propose the use of these images as hard-negatives samples, within the framework of hard-negative mining, in order to improve the performance of classification networks in applications which suffer from sparse labelled training data. Our approach makes use of a variational autoencoder (VAE) which is trained in an adversarial manner in order to learn a latent distribution of the training data, as well as to be able to generate realistic, high quality image patches. We evaluate our proposed generative approach to hard-negative mining on a synthetic aperture radar (SAR) and optical image matching task. Using an existing SAR-optical matching network as the basis for our investigation, we compare the performance of the matching network trained using our approach to the baseline method, as well as to two other hard-negative mining methods. Our proposed generative architecture is able to generate realistic, very high resolution (VHR) SAR image patches which are almost indistinguishable from real imagery. Furthermore, using the patches as hard-negative samples, we are able to improve the overall accuracy, and significantly decrease the false positive rate of the SAR-optical matching task—thus validating our generative hard-negative mining approaches’ applicability to improve training in data sparse applications.

Vision Based Approach to Sign Language Recognition

We propose an algorithm for automatically recognizing some certain amount of gestures from hand movements to help deaf and dumb and hard hearing people. Hand gesture recognition is quite a challenging problem in its form. We have considered a fixed set of manual commands and a specific environment, and develop a effective, procedure for gesture recognition. Our approach contains steps for segmenting the hand region, locating the fingers, and finally classifying the gesture which in general terms means detecting, tracking and recognising. The algorithm is non-changing to rotations, translations and scale of the hand. We will be demonstrating the effectiveness of the technique on real imagery.

Vision-based dynamic target trajectory and ego-motion estimation using incremental light bundle adjustment

This paper presents a vision-based, computationally efficient method for simultaneous robot motion estimation and dynamic target tracking while operating in GPS-denied unknown or uncertain environments. While numerous vision-based approaches are able to achieve simultaneous ego-motion estimation along with detection and tracking of moving objects, many of them require performing a bundle adjustment optimization, which involves the estimation of the 3D points observed in the process. One of the main concerns in robotics applications is the computational effort required to sustain extended operation. Considering applications for which the primary interest is highly accurate online navigation rather than mapping, the number of involved variables can be considerably reduced by avoiding the explicit 3D structure reconstruction and consequently save processing time. We take advantage of the light bundle adjustment method, which allows for ego-motion calculation without the need for 3D points online reconstruction, and thus, to significantly reduce computational time compared to bundle adjustment. The proposed method integrates the target tracking problem into the light bundle adjustment framework, yielding a simultaneous ego-motion estimation and tracking process, in which the target is the only explicitly online reconstructed 3D point. Our approach is compared to bundle adjustment with target tracking in terms of accuracy and computational complexity, using simulated aerial scenarios and real-imagery experiments.

ACTIVE SEGMENTATION

The human visual system observes and understands a scene/image by making a series of fixations. Every "fixation point" lies inside a particular region of arbitrary shape and size in the scene which can either be an object or just a part of it. We define as a basic segmentation problem the task of segmenting that region containing the "fixation point". Segmenting the region containing the fixation is equivalent to finding the enclosing contour — a connected set of boundary edge fragments in the edge map of the scene — around the fixation. This enclosing contour should be a depth boundary. We present here a novel algorithm that finds this bounding contour and achieves the segmentation of one object, given the fixation. The proposed segmentation framework combines monocular cues (color/intensity/texture) with stereo and/or motion, in a cue independent manner. The semantic robots of the immediate future will be able to use this algorithm to automatically find objects in any environment. The capability of automatically segmenting objects in their visual field can bring the visual processing to the next level. Our approach is different from current approaches. While existing work attempts to segment the whole scene at once into many areas, we segment only one image region, specifically the one containing the fixation point. Experiments with real imagery collected by our active robot and from the known databases1 demonstrate the promise of the approach.

The Implications of Image Collimation for Flight Simulator Training

There is some question as to whether non-collimated (i.e., real) imagery viewed at one meter or less provides sufficiently realistic visual cues to support out-the-window flight simulator training. As a first step toward answering this question, we have obtained perceived size and velocity estimates using both simple stimuli in a controlled laboratory setting and full simulator imagery in an apparatus consisting of optically combined collimated and real-image displays. In the size study it was found that real imagery appeared 15-30% smaller than collimated imagery. In the velocity studies, the laboratory data showed that the perceived velocity of real imagery was less than that of collimated imagery. No perceived velocity effects were found with the simulator imagery. Results support the position that for training tasks requiring accurate perception of spatial and temporal aspects of the simulated visual environment, misperceptions of size, but not velocity, need to be considered when real-image displays are used.