This week, class was cancelled so that each team can meet up and continue working on their project. The team met up and discussed the development of our AI model and the construction of our backend system. We took the time to figure out a different way to gather image data by fetching them from the Fanatics website directly. To do this, we would be using Python and its respective libraries to web scrape the website and pull both the PNG image and the JSON elements it has attached with it.

We also met with our liaison for the first time this semester and we discussed a summary of what the project is and how we are going to go about solving it. We also went through the entire QRB 1 with him and created a Slack channel so it can be easier to reach him for any questions.

Next week, we will be getting all the image data we need from the liaison, setting up the training on HiPerGator, continuing FDR development (Volume 1), and RSVPing for the FDR.

This week, we started class by reviewing our QRB 1 and speaking about the different challenges, positives, and negatives that it brought with it. This was done to improve our presentation for QRB 2 which we will be presenting in the next few weeks. We also took some time to work on our Final Design Report Outline, which comprises of all of the work that the team accomplished and will accomplish by the end of the semester.

Our team also took the time this week to complete our HiPerGator training, create a backend script to convert images into PNG, and develop a program to train our image generator (which still needs to categorize images appropriately).

Next week, we will be starting training on the HiperGator, finish coding to organize the backend data, and get more Fanatics data from the liaison.

This week, the Fantastic 5 presented to three judges, including a surprise guest, during our Qualitative Review Board. The team presented our project risks, project plan, and work distribution among the team. During our presentation, our coach Dr. Thomas brought in a surprise guest, the president of Palm University College in Ghana, Dr. Okantey! It was valuable presenting to him and getting his feedback. He found the work we were doing interesting and liked the progress we have made.

The main feedback our team got was to describe the AI metrics that are needed, such as sampling steps and noise level. We also received feedback about getting in touch with our liaison, as ever since he got reassigned, we have not heard back from our new one. We ended up contacting the liaison again and he said in the next two weeks they will reach out.

Next week, we are going to continue working on training the AI and developing the backend.

This week, the team had our first SCRUM meeting where we delegated tasks to each other and came up with our first sprint. Our sprints will last two weeks, and by the end of this sprint the AI model team will have figured out a way to start training the AI model using python. The backend team will focus on converting fanatics images to png and stripping attributes of the jerseys from a JSON.

This week, we also are preparing for our Qualitative Review Board coming up on Tuesday. We just have to finish up the PowerPoint and practice the presentation.

Next week, aside from presenting the QRB, we are hoping we hear back from our new liaison so we can set up a meeting as soon as possible. We need to get a bigger dataset from Fanatics, and we set up HiPerGator training for the upcoming Monday.

This week, the team started prepping for the Qualitative Review Board. We came up with our accomplishments last semester and what we have to do for the rest of the semester to meet our goals. Next, we will have to create a PowerPoint presentation to present it to the coaches on January 30th.

We also received sad news that our liaison last semester Jason got reassigned to another division within Fanatics so he will not be our liaison anymore. The team is sad to see him go, we enjoyed working with him. We have yet to meet our new liaison, but we are communicating with him to set up an appointment time.

The team also decided to split up the work of the project. Samy, Hadi and Joseph are going to work on the AI model and building it. While Sara and Aditi are going to work on the API to connect stable diffusion to Fanatics side code.

We are planning on following this guide, which shows big improvements to their prompt generation when a base stable diffusion model is tuned to the type of images they are looking for. If I follow this guide, i could improve the output we get from stable diffusion which generally is not great.

Hope everyone had a happy holiday season! The Fantastic 5 certainly did.

With the start of the Spring semester, comes the return of IPPD. Already on the first week of classes, the team dealt with a tornado warning and getting soaked from a thunderstorm heading to class. Classes were not cancelled unfortunately. One of our members was brave enough to face the storm head on. She is now sick.

Regardless, this week the Fantastic 5 completed a critical path plan for the rest of the semester to highlight major tasks that need to be achieved for success in the project. We also started correspondence involved with getting training for HiPerGator, as the team needs the supercomputer to train the AI model as it requires a lot of resources. We also had to complete a Work Breakdown Structure for January to distribute work for the month. This allows us to be more efficient and get started with our AI training.

Our Work Breakdown Structure is as follows for the month of January:

• 0.0 Product Image Generator for Fanatics Website
• 1.0 Project Management
• 2.0 Further Research
• 3.0 Stable Diffusion Development
  • 3.1 Testing
    • 3.1.1 Unit Tests
  • 3.2 Backend
    • 3.2.1 Train AI model using Stable Diffusion
    • 3.2.2 Integration with Fanatics Platform
    • 3.2.3 Stable Diffusion Handler
• 4.0 Qualification Review Board
  • 4.1 Create PowerPoint
  • 4.2 Rehearse Presentation

Next week, we are going to start implementing the AI model and preparing for the QRB.