Faire Home

Projects

Prizes

Directions

Faire Map


Spring 2007 Senior Projects

Click here for past projects

CS194 Projects


Adobe ASM - Andy Grant
ASM (Attack Surface Measurement) is a security tool designed up help software developers insure that their product is secure. ASM lists all the files a selected program accesses/creates and reports on the security attributes of these files. To do this, ASM attaches to a given process using system hooks and injects two DLLs. Now when ever the process attempts to read/write/create and file it is reported in a log. After reporting it in the log, ASM checks the properties of the file and prints out the security related attributes. ASM only runs on Windows.
 
Beer Pong - Ned Rockson, Luiz Pereira, Fred Thompson
Beer pong, also known as Beirut, is celebrated as one of the favorite pastimes in college campuses throughout the United States. However, there have been no real attempts to simulate the game, mostly because no human interface device was available to realistically capture its throwing motions. Enter the Nintendo Wii and its revolutionary controller with six degrees of motion sensing. Using the Wii controller as an input device, we have created Beer Pong: the first ever realistic Beirut simulator for the PC. Players take turns using the Wiimote to throw a virtual ping pong ball, in an attempt to land the ball in the opposing team's cups. Featuring realistic physics and polished graphics, players can pit themselves against each other or try their luck against varying levels of AI. Beer Pong's intuitive gameplay is fun and addictive for current Beirut players while allowing novice players a gentle introduction to the game. We hope you will join us for a new Beirut experience.
 
Blackhat - Victor Cepeda, Kurt Loidl, Benjie Nelson, Tom Wang
blackhat is a fast-paced 2 to 4 player action-arcade game developed for the XBox 360 and the PC that combines classical stylings of retro gaming with modern graphical flare to create a raucous good time. Players assume the role of cyber-hip hackers, using each character's unique digital avatar to vie for control of the Internet. blackhat features 4 unique hackers with a total of 13 different abilities and a rich digital environment that is never the same twice. Multiple game types, like Terminal Capture and Password Crack, mean limitless fun, and the simplicity of the game means that your little sister and your grandmother will make great hackers, too. blackhat is a game for anyone who has ever dreamt of being a superstar hacker.
 
Car 2.0 - Aaron Sullivan, Jordan Moncharmont
The Car 2.0 Platform allows for easy development of in-car Human Machine Interfaces (HMIs) or vehicle applications using standard web development languages (HTML, JavaScript, CSS, XML, etc.). Vehicle specific information (such as speed, location, input devices, etc.) is available via a JavaScript object, enabling the development of these interfaces and applications. Previous restrictions on supported OS's and browsers have been removed, making the Car 2.0 Platform work on any WiFi enabled device with a standard JavaScript-enabled browser.
 
Chummi - Jean Yang, Jennifer Gee, Raylene Yung
For college students who use Facebook, Chummi is a website that calculates compatibility between people, allowing users to discover new friendships as well as strengthen existing ones. Unlike matching sites such as eHarmony.com, our metrics do not require users to fill out extensive personality surveys, as we can leverage the wealth of untapped profile data on Facebook. The accuracy of our predictions is constantly improved as our algorithm is dynamically trained by user actiivity on the site. As an added bonus, we also provide trends on how students at our school select friends by analyzing aggregate data.
 
CourseRank - Benjamin Bercovitz, Filip Kaliszan, Henry Liou
CourseRank is an online peer-review system designed for students to share their feedback about different courses offered at their university. This service features grade distributions and ratings for every course and instructor, user comments with a reputation system, smart semantic search, and interesting statistics. As an advanced feature, CourseRank analyzes each user's rating history and compares it with all other users to provide relevant recommendations of classes the user might be interested in taking. All this is encapsulated in a friendly and intuitive user interface and has already proven to be an extremely useful tool to college students. CourseRank accumulated over 200 users and 1000 individual course ratings in its first week of beta release.
 
Exploring English - Trung Minh Lai, Matthew Brown, Joe Chen
Please find below for our project description: Exploring English is an interactive typing and writing learning application for the target users of the "Children's Machine" ($100 laptop). The One Laptop Per Child (OLPC) Typewriter application delivers an interactive keyboard typing and English language curriculum. To increase the speed, efficiency, and effectiveness of the user's typing skills, it emphasizes simultaneous typing and language skill development, utilizing an adaptive rhythm video game to maximize fun and engagement.
 
FutureForm - Amal Dar Aziz
FutureForm is a cross-disciplinary collaboration between Mechanical Engineering graduate students enrolled in ME310, students at St.Gallen University in Switzerland, and Computer Science senior Amal Dar Aziz. Sponsored by Deutsche Bahn, a premier German railway company, the goal of FutureForm is to equip future knowledge workers with the tools needed to collaborate effectively with remote co-workers, and work in flexible situations, without begin tied to a physical environment. This concept was realized through the creation of a dual-screen laptop. Through the use of the dual-screen laptop -- which allows for various configuration modes, to optimize working environments in both individual and collaborative work situations -- and software designed specifically for collaborative communication, including messenging software, virtual office tools, and a whiteboard application, among other softwares, the future knowledge can remain connected with their workplace while working away from home, a trend that is quickly increasing in popularity among the tech field. FutureForm truly adapts itself to the needs of every mobile worker, facilitating smoother and effective collaborations.
 
Homeslyce - Kavin Asavanant, Nate Janewit, Ellis Lau, Joyce Pan
Homeslyce is an online gift and money sharing service where users can purchase gifts for friends, dorms, couples, or any event! Users can find gifts on our service and "slice" them, paying only for their share of the gift. The store provides a complete experience, allowing users to navigate and find any gifts from Amazon or our own inventory and even obtain gift recommendations for oneself or for the recipient of the gift from their Facebook profile. Homeslyce will make the social troubles of gift-giving a distant memory.
 
IdeaCV - Kimber Lockhart, Rebecca Maxine Illowsky, Jeff Seibert
Think you have good ideas? IdeaCV is where you can show the world what you're all about. Get your name out, your insights, your knowledge. IdeaCV is at the forefront of the growing movement towards the open exchange and collaborative development of ideas. Don't miss out on the opportunity to establish your personal "brand" among the world's greatest innovators. Join forces with other thought-leaders, get feedback on projects, and discover trends before they develop. IdeaCV's innovative algorithms continually search the site and introduce you to others with similar ideas and interests. Build your portfolio now: the idea revolution has begun!
 
InteliSport - Amanda Luther, Rajesh Ranganath, Kyle Vigen
IntelliSport wraps an understandable user interface around Support Vector Machine artificial intelligence technology to allow the average sports fan to harness this powerful technique for classifying data and making predictions. For instance, given past data, users can predict who will win the 2007 NBA championship or determine which current NBA players are most likely to make the Hall of Fame. IntelliSport provides large datasets that help answer these questions, but users are also free to input their own data to make other predictions. Additionally, users can determine what the best predictive factors are, which allows them to evaluate the truth of sports clichés, such as the claim that defense wins championships.
 
A Multi-Touch Surface Using Multiple Cameras - Kevin Gabayan, Itai Katz
As interfaces increasingly become 3D, input devices with more expressivity than the keyboard and mouse are becoming necessary. In this project we present the framework for a gesture recognition-enabled workspace with a touch-sensitive desktop. With an overhead camera we can determine the x-y coordinates of the fingertips. With the addition of a side-mounted camera, we can detect touch events, as well as determine the z coordinate of the fingertip, yielding 3D localization with two low-cost webcams.
 
myFlee - Bhargav Raman, Akintunde Ismail Maiyegun, Michael Pabst
myFlee is a next generation, easy to use, site for college students to buy and sell. By letting myFlee users restrict their audience to their own campus, they can combine the convenience of an online auction site with the trust of a face-to-face exchange. myFlee users can post not only conventional auctions, Dutch auctions, and fixed-price sales, but can also negotiate loans and swaps of money, goods, and services. myFlee aims to provide every convenience for users, and integrates fleeMaps to setup meeting locations.
 
NavSync - Chris Balane, Tim Carter, Chris Collette, Chalmers Wang
NavSync is a project leveraging the BMW Car 2.0 platform. Our application takes advantage of the connection the Car 2.0 project can create between one's car and one's digital data. We created an in-car mapping navigational system which can harness the power of web-based API's for information aggregation and display while at the same time coupling this nav system with a home-computer based application that lets the user synchronize their daily agenda with the car. Taking advantage of the fact that an increasing number of people keep their schedules electronically, our Microsoft Outlook plugin uploads agenda data to a server-side application which stores the data and automatically configures the navigation system according to the driver's most-likely next destination. From there, our nav system utilizes car data to negotiate with web based mapping API's (i.e. Google Maps). Features built off of this include real time local gas pricing and satellite mapping. Our application demonstrates how a car's vital statistics can tailor the data specifically to the driver's needs and positions.
 
Peregrine Paths - Tony Cassanego, Dave Koslow, Ray Thang
Have you ever come back from a run curious to know exactly how far you went? Are you tired of going out for the same run or jog all the time? Do you ever feel like you have run every possible route in your area? If so, Peregrine Paths will be of help. Users can enter exercise routes in an easy-to-use Google Maps-based interface and determine both the length of their routes as well as the topographic information. Users can also browse routes anyone else has taken so that they can quickly find fresh ideas for where to exercise next. If a user cannot find a route that fits his or her needs, Peregrine Paths will generate a route using information from all paths in the area. Lastly, after doing all that exercise, users can easily track their progress by entering their times into Peregrine Paths after each workout. Feel free to head over to www.peregrinepaths.com before heading out for your next run!
 
RelatesTo - Jason Barrett Prado, Ben Trombley-Shapiro
Relates2 is a web-based application whereby users can declare a specific relationship between any two web addresses (URIs). Once the relationship between two sites is known, linked content can either be embedded throughout the web through our API or can be viewed using our Firefox plugin. In the end, users will enjoy a richer browsing experience that is augmented by relevant content largely from user-generated sources.
 
SafeSense - Belinda Gu, Nathan Marz, Ryan Propper, Chase Yarbrough
SafeSense is an extensible, in-car software system allowing real-time reporting of driver safety on the road. Developed in collaboration with BMW, SafeSense can, with user consent, report risk factors such as speeding, driving in high-risk areas, and driving during high-risk hours. As an application of this technology, the data collected can be used to reward statistically "safe" drivers with insurance discounts. SafeSense also features a novel "bad driver button," which enables users to report incidences of dangerous driving witnessed on the road in real-time. SafeSense uses sophisticated analysis to analyze this information, filter out dishonest reports, and determine the safety of each driver.
 
Sociophile - Ryan Fortune, Darius Henderson, Alex Medearis, Edmond Yap
Sociophile is a new way of sharing files. Unlike earlier models of sharing which leaves the user in a large pool of anonymous users, Sociophile allows users increased adminstration and selectivity of share peers and, thus, improved privacy and security. Furthermore, with groups centered around shared interests, Sociophile encourages community building, making the sharing experience even more enjoyable.
 
TypeTown Revolution - Nam Bui, Danny Bauman
Type Town Revolution is a typing game where players type a sequence of words in rhythm to a song (much like DDR). This game represents a departure from normal typing games because instead of focusing on the "type as fast as you can" aspect, it requires user to coordinate their typing with a beat. In addition to a normal scoring mode, there will also be a multiplayer mode where one user controls the left hand and the other controls the right, and a custom create-a-level mode.
 
Yay Gifts - Jeffrey Spehar, Bear Travis, Kalvin Wang, Tim Springer, Trung Vu
Many gifts are worthless to the recipient-- it's the thought (and money spent) that counts. So why not use the money for good? Yay Gifts lets people choose, pay for, and send virtual gifts; the proceeds go to charity. It's also clear from existing social networks and virtual worlds that people value possessions and enjoy public validation. We create an ecosystem of user-generated gifts and an alternative to buying trinkets to express appreciation or interest. Yay Gifts: doing good by changing gift-giving, one (virtual) gift at a time.
 

CS191W Projects

GenViewer - Jason Turner-Maier
We have sequenced the human genome, but do we understand how genetic differences within that genome affect factors such as genetic diseases and drug susceptibility? Genomica is a tool developed in Professor Daphne Koller's lab which allows scientists to use genomic and genetic expression data to learn relationships between genetic variation and such phenotypic attributes. However, it does not provide a convenient way to view the resulting models, limiting its potential impact. GenViewer is a Java-based visualization tool designed to be used with Genomica to view the learned relationships among genotypes, expression levels, and phenotypes. It supplies users with an easy way to explore these models and to understand the connections between genotypes and phenotypes.
 
Hopscotch - Chiedozi Acholonu
With the growing popularity of social networking, users have increasingly made available a signficant amount of personal information on numerous networking sites, such as MySpace and Facebook. In today's mobile, inter-connected world, it would be useful if users could access all of this information simultaneously on the go. Hopscotch is a new mobile user interface designed for the use of aggregating and viewing data from various social networks into one interface. The UI uses Adobe's Flash Lite to create an intuitive and visually pleasing experience that allows users to quickly and efficiently interact with profiles and photos on a mobile device. This proof of concept uses the Facebook API to exemplify the interaction model.
 
Multi-class image segmentation - David Cohen
Multi-class image segmentation, the task of automatically labeling each pixel in an image according to what real-world entity it depicts, has made significant advances in recent years through the combination of powerful machine learning techniques with rich features. Choices such as which types of features to use, how to construct them, and at what level of granularity to work can have dramatic effects on the success of an image segmentation system. In systems with multiple learning steps, the question of whether to add structure to the training data also becomes important. This paper provides insight into some of these choices through empirical comparisons of their effects on a state-of-the-art multi-class image segmentation system which uses simple appearance features and relative location information. We evaluate our results on the MSRC database and show that adding textons and maintaining a unified training set yield noteworthy improvements to our system, when and only when they are used in conjunction with features based on relative location information.
 

 



Faire Home | Projects | Prizes | Campus Map | Faire Map

Robert Plummer
plummer@cs.stanford.edu