Unsupervised Image-Generation Enhanced Adaptation for Object Detection in Thermal Images

Object detection in thermal images is an important computer vision task and has many applications such as unmanned vehicles, robotics, surveillance, and night vision. Deep learning-based detectors have achieved major progress, which usually need large amount of labelled training data. However, labelled data for object detection in thermal images is scarce and expensive to collect. How to take advantage of the large number labelled visible images and adapt them into thermal image domain is expected to solve. This paper proposes an unsupervised image-generation enhanced adaptation method for object detection in thermal images. To reduce the gap between visible domain and thermal domain, the proposed method manages to generate simulated fake thermal images that are similar to the target images and preserves the annotation information of the visible source domain. The image generation includes a CycleGAN-based image-to-image translation and an intensity inversion transformation. Generated fake thermal images are used as renewed source domain, and then the off-the-shelf domain adaptive faster RCNN is utilized to reduce the gap between the generated intermediate domain and the thermal target domain. Experiments demonstrate the effectiveness and superiority of the proposed method.

A Pseudo S-Plane Mapping of Z-Plane Root Locus

In this paper, inspired by the geometric inversion transformation, a novel transformation of the z-plane root locus to a pseudo s-plane is proposed. In the z-plane, the stability of a discrete closed-loop system (a sampled-data control system) requires that all the system poles lie within the unit circle. In root locus analysis, the unit circle region seems congested, compared to the stability region of a continuous system, which is the left half of the s-plane. In the case of fast sampling, the poles of a discrete system can really be in a small neighborhood, thus making the control implementation difficult. The geometric transformation developed in this work helps widen or enlarge the space for the system poles and preserves most of the features of z-plane root loci, including marginal stability and root loci branching off at vertical angles. The usefulness of the new transformation in design of discrete control systems is demonstrated in a numerical example.

Uniqueness of meromorphic functions concerning their difference operators and derivatives

Let f and g be two nonconstant meromorphic functions. Shared value problems related to f and g are investigated in this paper. We give sufficient conditions in terms of weighted value sharing which imply that f is a linear transformation or inversion transformation of g. We also investigate the uniqueness problem of meromorphic functions with their difference operators and derivatives sharing some values.

Analytical Solution for the Steady-State Karst Water Inflow into a Tunnel

An analytical solution for the karst water inflow into a lined tunnel in an infinite plane is derived based on conformal mapping. The new solution considers the center distance between the tunnel and the cavern, the radii of the tunnel and the cavern, and the property of the lining, such as the permeability coefficient as well as the lining radius. Numerical models are established and calculated using the finite difference software FLAC3D to compare with the analytical solution of inversion transformation, and a good agreement is found. Then, the parameters of effect are discussed in detail. The results indicate that the karst water inflow shows a curve relationship as the radius of tunnel increase and increases as the lining becomes thinner or the permeability coefficient of the lining increases. Moreover, the pressure head decreases as the tunnel radius and the center distance between the tunnel and the cavern increase.

STAR-SHAPED SET INVERSION FRACTALS

In the paper, we generalized the idea of circle inversion to star-shaped sets and used the generalized inversion to replace the circle inversion transformation in the algorithm for the generation of the circle inversion fractals. In this way, we obtained the star-shaped set inversion fractals. The examples that we have presented show that we were able to obtain very diverse fractal patterns by using the proposed extension and that these patterns are different from those obtained with the circle inversion method. Moreover, because circles are star-shaped sets, the proposed generalization allows us to deform the circle inversion fractals in a very easy and intuitive way.