This week I worked on installing openpose on an amazon ec2 instance. Along the way I found I needed to use a p2 type of server that would have the required tools for openpose. I didn’t initially have access to this so I requested it and have been approved but haven’t finished this up yet.
I also worked with the library and got json data for 3 basic workout images, one for each exercise. I then analyzed the data to get a better understanding of how we will work with this data, and how we will communicate it across our cloud services. This is important to be confident about because this data is essential to our classifier, rep/set algorithm, and our form correction algorithm.
In the next stretch of time I will install the openpose library on a p2 instance and begin to test it to see the performance and decide if we need a more intensive server. Then I will integrate AWSÂ Kinesis to get frame data on the AWS server that is running openpose.
I will work with Cooper to standardize the output of the hardware camera so that we can begin to collect training data for our classifier. I will work with Nakul to begin designing our backend endpoints.
I think we are still on progress this week as long as I do not run into any issues with openpose on the p2 instance, since that is what caused a lag this week.
