Search

Accurately estimating the position of static objects, such as traffic lights, from the moving camera of a self-driving car is a challenging problem. In this work, we present a system that improves the localization of static objects by jointly-optimizing the components of the system via learning. Our system is comprised of networks that perform: 1) 5DoF object pose estimation from a single image, 2) association of objects between pairs of frames, and 3) multi-object tracking to produce the final geo-localization of the static objects within the scene. We evaluate our approach using a publicly-available data set, focusing on traffic lights due to data availability. For each component, we compare against contemporary alternatives and show significantly-improved performance. We also show that the end-to-end system performance is further improved via joint-training of the constituent models. Code is available at: https://github.com/MedChaabane/Static_Objects_Geolocalization.

Accurate and topical spatial datasets representing human populations are foundational to solving humanitarian issues. This paper provides unique solutions to accurately map human settlements both at scale and across remote areas. A method of village boundary extraction from very high resolution optical satellite imagery is proposed. Furthermore, the supplement of a crowd-sourced validation process to filter the detections for higher accuracy and the automated mosaic techniques are detailed. To demonstrate the computational and informational scalability of the process, four distinct geographic locations in Nigeria, Somalia, Pakistan, and Afghanistan were analyzed for a total processed area of 592 000 km 2 , comprised of 1159 high-resolution DigitalGlobe images. The geographic variability of the locations and the scale of the projects required dynamic and automated solutions, respectively. The curated results exhibit high recall and precision of human settlement data in remote as well as urban areas. Crowdsourced validation allows complete control over the precision of the final village boundary layer, and given time, an effective 100% precision can be achieved. This highly scalable and precise system is perfectly adequate for processing regional and country-scale areas, with minimal human effort.

Building footprints (BFP) provide useful visual context for users of digital maps when navigating in space. This paper proposes a method for extracting and symbolizing building footprints from satellite imagery using a convolutional neural network (CNN). The CNN architecture outputs rotated rectangles, providing a symbolized approximation that works well for small buildings. Experiments are conducted on the four cities in the DeepGlobe Challenge dataset (Las Vegas, Paris, Shanghai, Khartoum). Our method performs best on suburbs consisting of individual houses. These experiments show that either large buildings or buildings without clear delineation produce weaker results in terms of precision and recall.

This chapter presents a selection of image representation methods for very high spatial resolution optical data acquired by agile satellite platforms. Each method aims at specific properties of the image information content and can be tailored to address unique features in spatial, temporal, and angular acquisitions. Techniques for the identification and characterization of surface structures and objects often employ spatial and spectral features best represented in panchromatic and multi-spectral images, respectively. In both cases, the vastness of the data space can only be addressed effectively by means of some data representation structure that organizes the image information content in meaningful ways. The latter suggest that a globally optimal representation of the object(s) of interest can be obtained through interactions with a scale space as opposed to single-scale information layer. Two examples, the Max-Tree and Alpha-Tree algorithms, are discussed in the context of interactive big data information mining. Optical and structural properties of the surface materials can be exploited by analyzing the tempo-angular domain by means of anisotropic decompositions, which rely on the availability of surface reflectance data and dense angular sampling. Bidirectional reflectance distribution functions of various materials are discussed in detail, showing that the temporal information should always be coupled to the corresponding angular component to make the best use of the available imagery.

A method for analyzing the urbanization process from multitemporal SPOT 5 panchromatic images is presented. The analysis is performed by unsupervised change detection between built-up presence indicators extracted separately from the scenes. The obtained index has an effective resolution which is coarser than the Ground Sample Distance, thus allowing a good spatial match between the indicators. Then, a local Mixed Information change indicator is employed in order to capture the non-linear temporal behaviors. A region-based approach is developed in synergy with the local Mixed Information in order to process consistently large scenes which may be affected by regional acquisition distortions. Experiments are conducted on SPOT 5 scenes acquired in 2003 and 2008 which cover a suburban area (45×45 km2) of the city of Tangshan in China. The urbanization detector is validated by visual interpretation, giving an equal omission/commission probability of 10%. A comparison to linear change indicators highlights an improvement of 15% of the equal omission/commission probability.