CameramanVis : where the camera should look ? CPSC 547 Project proposal
نویسنده
چکیده
In the soccer broadcasting, human camera operators adjust the pan angle of a pan, tilt and zoom (PTZ) camera to follow player positions and ball position. In this project, we assume the camera only has 1 degree of freedom (DoF) which is the pan angle. And the player positions are the main targets when the cameraman shots the video. The project aims to analyze the relationship between the pan angle and player position. The dataset is from my research project. It was collected from a semi-professional soccer game. The dataset has 172, 800 frames (48 minutes). Each frame contains a set of player positions and a camera pan angle. The player position is measured on the playing ground and are quantized to a 14 dimension feature. For simplicity, we define following terms: • frame:one image in a video. • frame number: a non-negative integer. It is quantized time. • camera angle: camera pan angle. It is a scalar value in one frame. • player location: player locations on the playing ground in one frame. It is a set of 2D locations. • feature: quantized player locations in one frame. It is a 14 dimension vector. The quantization method is described in [3]. • sample: a pair of a camera angle and a set of player position that are captured (detected) at the same frame. • camera angle distribution: a histogram of a set of camera angles. • feature distribution: mean and standard deviation of features. We assume each dimension is independent. • feature distance: Euclidean distance between two features. To reduce the redundant information and make the program faster, we uniformly sampled data from the original data. To evaluate the information loss in sampling, we linsampling step 10 15 20 25 30 35 40 45 50 55 60 re co ns tru ct io n er ro r ( de gr ee /fr am e)
منابع مشابه
CameramanVis: where the camera should look? CPSC 547 Project update
In sports analysis, player trajectory and occupancy map are widely used for visualizing interesting objects (players and balls). Lines are overlaid on the court to visualize the trajectory of players. For example, Fig. 1 (a) shows the trajectory of team centroid. Also, the derived game phases (e.g. offensive play, defensive play and time out) from the trajectory are visualized by ellipses (Fig....
متن کاملCameramanVis: Where the Camera Should Look
This paper introduces CameramanVis, an interactive visualization system for visualizing correlation between camera angles and player locations for soccer games. The system supports computer vision researchers to browse the soccer data, to identify camera angle and player location patterns, and to identify and compare outliers. The system visualizes camera angle, player locations and their corre...
متن کاملVisualizing Trees of a Random Forest Model CPSC 547 Project Proposal
It is usually not difficult to visualize an entire classification tree in a single image without overwhelming the user with visual clutter. Interpretation becomes a problem when we are required to visualize thousands of trees at once. Random forests have become a very common data mining algorithm due to advantages of high accuracy and ability to handle large amounts of input variables. A random...
متن کاملGazeVis: An Eye-Tracking Visualization Towards Predicting User Distraction CPSC 547 Project Report
We introduce GazeVis, a tool to visualize data gathered from a PDF reader application called GazeReader developed in another course. Our visualization aims to understand a reader’s gaze pattern before a self interruption occurs. We achieve this by allowing users to interactively inspect gaze related features over time. Furthermore, our visualization incorporates a prediction that determines if ...
متن کاملذخیره در منابع من
با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید
عنوان ژورنال:
دوره شماره
صفحات -
تاریخ انتشار 2015