Feature data

Abstract: A geographical position/image capturing system stores object images and position coordinates as digital data. The system incorporates a geo-addressed map data base and geo-positioning device for relating the position of the system at the time the image is captured to the captured digital image data and the geo-addressed map. A point-of-interest feature data base is linked to the image data and the position data, by hyper-media links. Digital multi-media entities such as graphics, video clips, audio streams and the like can be digitally stored and retrieved based on hyper-media links coupling the entities to the object images, the map position and feature data base. A playback unit incorporating an image viewer communicates with the stored digital digital image data, the multi-media entities, the map data base, and the feature data base and allows modified images of selected portions of the images and other data to be viewed by a user. An audio reproduction device may be included to reproduce the audio media entities when they are selected by the appropriate hyper-media link. In one embodiment of the system, the image capturing function and the geo-positioning function are physically remote from the viewing function until the captured data is transferred to the retrieval and playback unit. In another embodiment, the capture functions and playback functions are combined in a compact, hand-held, portable unit for consumer or industrial field use. The images, position data and multi-media entities may be stored as a single, compressed, linked digital data file. The data file is provided with hyper-media connections between the various entities, which connections are actuated by clicking on icons displayed on the viewer, or by actuating buttons or switches provided on the system console. Angular orientation determining device for providing angular orientation data may also be incorporated such that the system angular orientation at the time of image capture may be captured and stored for subsequent retrieval and display.

Abstract: Recent studies have shown that the efficiency of deep neural networks in mobile applications can be significantly improved by distributing the computational workload between the mobile device and the cloud. This paradigm, termed collaborative intelligence, involves communicating feature data between the mobile and the cloud. The efficiency of such approach can be further improved by lossy compression of feature data, which has not been examined to date. In this work we focus on collaborative object detection and study the impact of both near-lossless and lossy compression of feature data on its accuracy. We also propose a strategy for improving the accuracy under lossy feature compression. Experiments indicate that using this strategy, the communication overhead can be reduced by up to 70% without sacrificing accuracy.

Abstract: The authors present 3D-POLY, a working system for recognizing objects in the presence of occlusion and against cluttered backgrounds The time complexity of this system is only O(n/sup 2/) for single-object recognition, where n is the number of features on the project The organisation of the feature data for the models is based on a data structure called the feature sphere Efficient constant-time algorithms for assigning a feature to its proper place on a feature sphere and for extracting the neighbors of a given feature from the feature sphere representation are presented For hypothesis generation, local feature sets are used The combination of the feature sphere idea for streamlining verification and the local feature sets for hypothesis generation results in a system whose time complexity has a low-order polynomial bound >