Detecting and reading text from natural images is a hard computer vision task that is central to a variety of emerging applications. Related problems like document character recognition have been widely studied by computer vision and machine learning researchers and are virtually solved for practical applications like reading handwritten digits. Reliably recognizing characters in more complex scenes like photographs, however, is far more difficult: the best existing methods lag well behind human performance on the same tasks. In this paper we attack the problem of recognizing digits in a real application using unsupervised feature learning methods: reading house numbers from street level photos. To this end, we introduce a new benchmark dataset for research use containing over 600,000 labeled digits cropped from Street View images. We then demonstrate the difficulty of recognizing these digits when the problem is approached with hand-designed features. Finally, we employ variants of two recently proposed unsupervised feature learning methods and find that they are convincingly superior on our benchmarks.

/pdf/reading-digits-in-natural-images-with-unsupervised-feature-3ckuq1665o.pdf

Reading Digits in Natural Images with Unsupervised Feature Learning

From the Publisher: 
The accessible presentation of this book gives both a general view of the entire computer vision enterprise and also offers sufficient detail to be able to build useful applications. Users learn techniques that have proven to be useful by first-hand experience and a wide range of mathematical methods. A CD-ROM with every copy of the text contains source code for programming practice, color images, and illustrative movies. Comprehensive and up-to-date, this book includes essential topics that either reflect practical significance or are of theoretical importance. Topics are discussed in substantial and increasing depth. Application surveys describe numerous important application areas such as image based rendering and digital libraries. Many important algorithms broken down and illustrated in pseudo code. Appropriate for use by engineers as a comprehensive reference to the computer vision enterprise.

Computer vision : a modern approach = 计算机视觉 : 一种现代的方法

IEEE Transactions on Pattern Analysis and Machine Intelligence

/pdf/a-break-in-the-clouds-towards-a-cloud-definition-svovsw1r5a.pdf

A Break in the Clouds: Towards a Cloud Definition

Robust detection and tracking of objects is crucial for the deployment of autonomous vehicle technology. Image based benchmark datasets have driven development in computer vision tasks such as object detection, tracking and segmentation of agents in the environment. Most autonomous vehicles, however, carry a combination of cameras and range sensors such as lidar and radar. As machine learning based methods for detection and tracking become more prevalent, there is a need to train and evaluate such methods on datasets containing range sensor data along with images. In this work we present nuTonomy scenes (nuScenes), the first dataset to carry the full autonomous vehicle sensor suite: 6 cameras, 5 radars and 1 lidar, all with full 360 degree field of view. nuScenes comprises 1000 scenes, each 20s long and fully annotated with 3D bounding boxes for 23 classes and 8 attributes. It has 7x as many annotations and 100x as many images as the pioneering KITTI dataset. We define novel 3D detection and tracking metrics. We also provide careful dataset analysis as well as baselines for lidar and image based detection and tracking. Data, development kit and more information are available online.

nuScenes: A multimodal dataset for autonomous driving

With the growth of LoRa deployments there are plenty of anecdotal reports of very long wireless links, well beyond the line of sight. Most reports suggest that these links are related to anomalous tropospheric propagation. We developed a platform to study tropospheric links based on TheThingsNetwork, a popular LoRaWAN-based infrastructure. We present some preliminary results and call for the IoT community to participate in this radio propagation experiment.

TROPPO LoRa: TROPospheric Personal Observatory using LoRa signals

Providing Internet connectivity to schools has been identified as paramount for development, for instance in the Giga project, cosponsored by ITU and UNICEF, with the goal of connecting every school to the Internet by 2030. For a country wide deployment, it is imperative to perform a thorough planning of the whole installation, using radio frequency (RF) propagation models. While statistical models based on empirical RF propagation data gathered in different scenarios can be employed, for point to point links at microwave frequencies the existence of a clear line of sight (LOS) is normally a prerequisite. The Irregular terrain model which makes use of digital elevation maps (DEM) has proved quite effective for simulating point to point links, but its application to a great number of links becomes time consuming, so we have developed an automated framework to perform this task. As a case study we have applied this framework in the planning of a project mired at providing connectivity to primary and secondary schools all over the country of Malawi.

Massive RF Simulation Applied to School Connectivity in Malawi

Localization in Smart Applications

Despite tremendous growth in the area of mobile telephony, Africa continues to lag behind other developing regions in Internet access. The business opportunity to capitalize upon this Internet Protocol (IP) traffic need has not gone unnoticed. Major communications service providers have been investing heavily in Ka-band communications (the Ka-band is part of the K band of the electromagnetic spectrum) — which can carry more data than lower frequencies and offers additional frequency ranges at already occupied satellite positions — such that next generation Ka-band satellites can be responsive to the burgeoning IP traffic market. However, these Kaband systems have disadvantages compared to lower frequency solutions; Ka-band systems are much more susceptible to weather due to signal absorption by moisture in the air and by wetness on antenna surfaces. These inherent limitations of the Ka-band hold a special significance for communications in Africa since almost a third of the continent is tropical in climate, and weather patterns across the continent are expected to further increase in variability, including increases in the frequency of extreme events, such as storms. The Paper further explores how the use of TV White Spaces, in addition to the availability of real-time weather information via the notion of “humans of sensors,” could be particularly useful for observing, anticipating, and mitigating disruptions to communications services.

/pdf/mitigation-of-rain-induced-ka-band-attenuation-and-29uecxp7nd.pdf

Marco Zennaro

Papers

TROPPO LoRa: TROPospheric Personal Observatory using LoRa signals

Massive RF Simulation Applied to School Connectivity in Malawi

Localization in Smart Applications

Mitigation of Rain-Induced Ka-Band Attenuation and Enhancement of Communications Resiliency in Sub-Saharan Africa

Risks Reduction of Rainfall-Induced Landslides-A Site-Specific Early Warning System (SSEWS)