There is often more than one way to select tonal adjustment-for a photograph, and different individuals may prefer different adjustments. However, selecting good adjustments is challenging. This paper describes a method to predict whether a given tonal rendition is acceptable for a photograph, which we use to characterize its range of acceptable adjustments. We gathered a dataset of image “acceptability” over brightness and contrast adjustments. We find that unacceptable renditions can be explained in terms of overexposure, under-exposure, and low contrast. Based on this observation, we propose a machine-learning algorithm to assess whether an adjusted photograph looks acceptable. We show that our algorithm can differentiate unsightly renditions from reasonable ones. Finally, we describe proof-of-concept applications that use our algorithm to guide the exploration of the possible tonal renditions of a photograph.

Adjusting photographs to obtain compelling renditions requires skill and time. Even contrast and brightness adjustments are challenging because they require taking into account the image content. Photographers are also known for having different retouching preferences. As the result of this complexity, rule-based, one-size-fits-all automatic techniques often fail. This problem can greatly benefit from supervised machine learning but the lack of training data has impeded work in this area. Our first contribution is the creation of a high-quality reference dataset. We collected 5,000 photos, manually annotated them, and hired 5 trained photographers to retouch each picture. The result is a collection of 5 sets of 5,000 example input-output pairs that enable supervised learning. We first use this dataset to predict a user's adjustment from a large training set. We then show that our dataset and features enable the accurate adjustment personalization using a carefully chosen set of training photos. Finally, we introduce difference learning: this method models and predicts difference between users. It frees the user from using predetermined photos for training. We show that difference learning enables accurate prediction using only a handful of examples.

Adjusting photographs to obtain compelling renditions requires skill and time. Even contrast and brightness adjustments are challenging because they require taking into account the image content. Photographers are also known for having different retouching preferences. As the result of this complexity, rule-based, one-size-fits-all automatic techniques often fail. This problem can greatly benefit from supervised machine learning but the lack of training data has impeded work in this area. Our first contribution is the creation of a high-quality reference dataset. We collected 5,000 photos, manually annotated them, and hired 5 trained photographers to retouch each picture. The result is a collection of 5 sets of 5,000 example input-output pairs that enable supervised learning. We first use this dataset to predict a user's adjustment from a large training set. We then show that our dataset and features enable the accurate adjustment personalization using a carefully chosen set of training photos. Finally, we introduce difference learning: this method models and predicts difference between users. It frees the user from using predetermined photos for training. We show that difference learning enables accurate prediction using only a handful of examples.

The impressive penetration of 802.11-based wireless networks in many metropolitan areas around the world offers, for the first time, the opportunity of a "grassroots" wireless Internet service provided by users who "open up" their 802.11 (Wi-Fi) access points in a controlled manner to mobile clients. While there are many business, legal, and policy issues to be ironed out for this vision to become reality, we are concerned in this paper with an important technical question surrounding such a system: can such an unplanned network service provide reasonable performance to network clients moving in cars at vehicular speeds.To answer this question, we present the results of a measurement study carried out over 290 "drive hours" over a few cars under typical driving conditions, in and around the Boston metropolitan area (some of our data also comes from a car in Seattle). With a simple caching optimization to speed-up IP address acquisition, we find that for our driving patterns the median duration of link-layer connectivity at vehicular speeds is 13 seconds, the median connection upload bandwidth is 30 KBytes/s, and that the mean duration between successful associations to APs is 75 seconds. We also find that connections are equally probable across a range of urban speeds (up to 60 km/hour in our measurements). Our end-to-end TCP upload experiments had a median throughput of about 30 KBytes/s, which is consistent with typical uplink speeds of home broadband links in the US. The median TCP connection is capable of uploading about 216 KBytes of data.Our high-level conclusion is that grassroots Wi-Fi networks are viable for a variety of applications, particularly ones that can tolerate intermittent connectivity. We discuss how our measurement results can improve transport protocols in such networks.

No abstract available.

CarTel is a mobile sensor computing system designed to collect, process, deliver, and visualize data from sensors located on mobile units such as automobiles. A CarTel node is a mobile embedded computer coupled to a set of sensors. Each node gathers and processes sensor readings locally before delivering them to a central portal, where the data is stored in a database for further analysis and visualization. In the automotive context, a variety of on-board and external sensors collect data as users drive.CarTel provides a simple query-oriented programming interface, handles large amounts of heterogeneous data from sensors, and handles intermittent and variable network connectivity. CarTel nodes rely primarily on opportunistic wireless (e.g., Wi-Fi, Bluetooth) connectivity to the Internet, or to "data mules" such as other CarTel nodes, mobile phone flash memories, or USB keys-to communicate with the portal. CarTel applications run on the portal, using a delay-tolerant continuous query processor, ICEDB, to specify how the mobile nodes should summarize, filter, and dynamically prioritize data. The portal and the mobile nodes use a delay-tolerant network stack, CafNet, to communicat.CarTel has been deployed on six cars, running on a small scale in Boston and Seattle for over a year. It has been used to analyze commute times, analyze metropolitan Wi-Fi deployments, and for automotive diagnostics.

We propose a reusable data management system, called CarTel, for querying and collecting data from intermittently connected devices. CarTel provides a simple, incrementally-deployable platform for developing automobile-based sensor applications. Our platform provides a dynamic query system that allows both continuous (standing) and one-shot geo-spatial queries over car position, speed, and sensory data as well as a both a low-cost/high-bandwidth substrate for communicating with a large network of mobile devices.

No abstract available.