Abstract: Many unsupervised domain adaptation (UDA) methods have been proposed to bridge the domain gap by utilizing domain invariant information. Most approaches have chosen depth as such information and achieved remarkable successes. Despite their effectiveness, using depth as domain invariant information in UDA tasks may lead to multiple issues, such as excessively high extraction costs and difficulties in achieving a reliable prediction quality. As a result, we introduce Edge Learning based Domain Adaptation (ELDA), a framework which incorporates edge information into its training process to serve as a type of domain invariant information. Our experiments quantitatively and qualitatively demonstrate that the incorporation of edge information is indeed beneficial and effective, and enables ELDA to outperform the contemporary state-of-the-art methods on two commonly adopted benchmarks for semantic segmentation based UDA tasks.

Abstract: A metal product includes a metal substrate, at least one first hole, at least one second hole, and at least one third hole. The first hole is formed in a surface of the metal substrate. The second hole is formed in at least one of a portion of the surface of the metal substrate without the first hole and an inner surface defining the first hole. The third hole is formed in at least one of a portion of the surface of the metal substrate without the first hole and without the second hole, a portion of the inner surface defining the first hole without the second hole, and an inner surface defining the second hole. The first, second, and third holes enhance a bonding strength between the metal product and a material product. The disclosure also provides a metal composite and a method for manufacturing the metal product.

Abstract: An audio signal processing method is provided. The method separates a left and a right channel audio signal to a center-left and center-right channel audio signals, and a side-left and side-right channel audio signals, and processes the center and side channel audio signals by a Head Related Transfer Function (HRTF) so as to simulate the audio signals into several audio source positions related to a user, then synthesizes the processed audio signals into dual channel audio signals according to the user's hearing characteristics by a frequency compensation process. The invention further provides an audio signal processing apparatus to perform the audio signal processing method.

Abstract: A mobile information server system is installed on mobile devices. When more than two mobile devices are carried by users, each of the mobile devices sends a geographic position and an IP address to a network location server to be stored and updated by the network location server. The network location server further analyzes the geographic position of each of the mobile devices, and sends the geographic positions and the IP addresses of other mobile devices that are nearby the mobile devices to each of the mobile devices, respectively. And then, the mobile devices obtain and display the information published by the other mobile devices through software protocol to accomplish the function of local business service. The network location server delivers the geographic position and the IP address, but not other information, and each of the mobile devices plays a role of an information server.