Sharp Shooting: Improving Basketball Shooting Form
Research on existing motion data collection software has been collected, and different software will be tested throughout this project. Materials, including the Kinect, were obtained. Initial motion data has been collected using a sample program written in Processing, and position data has been displayed in a software called Isadora.
Engineering Statement: Purpose: Background: Engineering Goal: The goal of this project is to engineer software which records data on basketball shooting form, analyzes the data, and gives accurate instructional feedback. Basketball is one of the most popular sports in the United States and more people are learning to play it around the world. Becoming a good basketball player and developing good shooting form requires lots of practice and feedback on how to improve. This feedback is often given by coaches and other basketball players, but other people are not always around to watch players practice. Because of this hindrance, there is a need for feedback on basketball shooting form which can always be given to its users.
Most feedback on basketball shooting form consists of qualitative observations from spectators. The lack of instructional feedback makes it difficult to learn proper basketball shooting form. Proper basketball shooting form is important to learn in order to make more shots. During a proper basketball shot, a player should balance with their feet shoulder width apart, knees bent, and back straight. The arm should then be raised above the head of the shooter with the elbow to the side of the shooter pointed toward the ground and the wrist should snap forward and down. There are many different predictions for the optimal angle for a basketball shot, but the majority of those predictions fall between 50 and 55 degrees. The variety of optimal angles for basketball shooting form suggest that there is a range of angles which will result in a made basket. Shooting form motion data can be collected in a variety of different ways including sensors (such as a Kinect) or video analysis. There is a need to improve on competing devices which analyze shooting form because of low accuracy, lack of availability, inability to predict and give feedback, and high cost.
Decision Matrix: Table 1. Basketball Shot Decision Matrix The type of basketball shot being analyzed needs to be consistent throughout data collection. Therefore, a decision had to be made on which kind of shot best fit the ideal specifications. Criteria Scale Free Throws Jumpshots Layups large range of shooting forms 3 5 5 1 worth the most points in a game 1 1 5 3 easy to record data 5 5 1 1 easy to replicate shot 3 5 1 5 Total: 65 30 29
Materials: MacBook Pro Microsoft Kinect Version One (Model 1414) USB AC adapter power supply cable USB to USB-C adapter Processing 2 Isadora
Procedure: Research on existing motion data collection software has been collected, and different software will be tested throughout this project. Materials, including the Kinect, were obtained. Initial motion data has been collected using a sample program written in Processing, and position data has been displayed in a software called Isadora. In order to start collecting motion data, the Kinect needs to be plugged in to the wall and to a computer. Then, a software which recognizes human joints from the infrared image detected by the Kinect must be run. Motion data must be recorded and compiled in a raw data file which is ready for analysis. As data analysis is being completed, instructional feedback must be created based on ideal joint velocities and angles.
Results: Initial position data has been collected from still images via Processing and Isadora. The three images are displayed below, along with the xyz coordinate data of each point:
Figure 1. Kinect Skeleton A Figure 2. Kinect Skeleton B
Body Part Position x y z head 104 455 2534 neck 91 239 2521 torso 78 26 2494 left shoulder -68 247 2532 left elbow -455 232 2558 left hand -782 230 2596 right shoulder 249 231 2511 right elbow 627 166 2500 right hand 955 107 2513 left hip -38-182 2473 left knee -81-595 2350 Figure 3. Kinect Skeleton C left foot -152-963 2314 right hip 170-193 2460 right knee 233-614 2384 right foot 334-976 2352
Body Part Position x y z head 102 272 2247 neck 96 62 2262 torso 93-143 2278 left shoulder -74 65 2262 left elbow -145-120 2112 left hand -66 174 2021 right shoulder 266 60 2261 right elbow 413-133 2196 right hand 247-50 2004 left hip -6-348 2295 left knee -142-640 2175 left foot -196-1019 2160 Table 2. Kinect Skeleton A Position Data Table 3. Kinect Skeleton B Position Data Table 4. Kinect Skeleton C Position Data right hip 187-351 2295 right knee 339-661 2115 right foot 308-1075 2295
Body Part Position x y z head 99 337 2261 neck 102 104 2256 torso 94-103 2275 left shoulder -44 108 2234 left elbow -66 248 1989 left hand -50 523 1808 right shoulder 249 100 2279 right elbow 422-86 2242 right hand 204 50 2101 left hip -10-308 2280 left knee -93-624 2180 left foot -171-877 2142 right hip 180-313 2309 right knee 340-634 2162 right foot 307-986 2335
Data Analysis: It is important to get quick data analysis and feedback from motion data in order to provide a user with feedback as they are practicing. Data should be analyzed using a program which calculates joint angles and velocities. This program should then compare this data to ideal measurements and determine how probable it is that the shot will go in the basket. Then, instructional feedback must be created based on the analysis of the motion data and given to the user.
Design Criteria: Future Work: The feedback given should do the following: o gives probability of shot falling o addresses a specific part of shot motion to change and how to change it The feedback should also meet the following criteria o highly accurate o highly precise o clear o audible o easy to visualize what needs to be changed o creates change in the shooting form of the user
The following pieces of my project are modular and can be completed in any order, but the following goals should be met before February Fair: Ideal form motion data must be measured and recorded A way to compare the shot of a user to the ideal form must be created Software must be created which predicts the chance of a shot falling and gives accurate, instructional feedback on how to fix shooting form Other goals have been established for long term research as well: Predictions and shot analysis can be incorporated into fast-paced basketball games and practices Ideal shooting forms of other people can be measured and compared Research can be applied to other fields which require motion data analysis such as other sports, physical therapy, and security
Timeline: December 20 January 3 January 31 data collection software should be chosen initial shooting motion data should be recorded initial data analysis should be completed sample feedback should be created more precise analysis and feedback should be completed software should be created feedback vocal recordings should be completed Figure 4. Project Timeline and Goals