I moved into campus and got set up with the windows computer on the lab. I installed the necessary apps and set up my working space. I finally started working through some photographs and Adobe Lightroom CC, and I finished the FASTAI lesson 1. I am planning on exploring the Unreal Engine simulation content this week by working through some tutorials while we wait for Reality Capture.
- Machine Learning:
- Rewatched some parts of Lesson 1 from the FASTAI tutorials. I have completed the tutorial in VSCode. I got an understanding of the training of models, and the usage of some pretrained models. I thought it was intersting how some pretrained networks could take seconds to train with a 99% rate of accuracy.
- I was able to get running a jupypter notebook on VSCode without problems, last week I had a hard time running Python code on the notebook since an older version was being used, I am now using Python 3.9 I believe.
- I went through the paper on "Deep Visual Domain Adaptation: A Survey" The paper was interesting, but I thought it was a bit dense and hard to follow. I had a hard time understanding some of the topics mentioned on the paper, so I am hoping to read on the CNN paper or look through other information on Domain Adaptation. It seems like Neural Networks are interesting, but there is a lot to unpack.
- Learned/ added aliases on bash to create shortcuts.
- Got VSCode running on the lab's computer
- Photogrammetry:
- Checked out the Nikon D3300 from ITS
- I shot 126 images of a stool from the Estella courtyard, lighting was a bit harsh since there were no white clouds. It was a bit difficult to get lower angle shots, but I was able to get layered photos from different positions.
- Corrected the exposure to some images. I corrected the highlights (lowered) and the shadows(increased). The images are flat and I believe they ready to be used in Reality Capture, subject to test when we get our hands on the software.
- Emailed Capture Reality to potentially get access to Reality Capture software. Liz filled out the form to hopefully get it for free since it is for academic purposes.
- Installed and set the computers with proper drivers(for NVIDIA GPU), Adobe Lightroom CC, Epic Games, Unreal Engine, and VSCode
- We have not been able to get access to the Reality Capture software.
- While I believe the images to be workable, I am aiming to retake some of them since some angles were hard to reach. The hot weather made it hard to get closer to the floor to take some of these images. I am hoping that with a little but more practice alongside with better weather or useful camera gadgets, we can get clear low angle pictures for a small model (the stool for us).
- Work on another lesson for fastai
- Read/ look into the "Visualizing and Understanding Convolutional Neural Networks" and/or "Investigating Neural Network Architectures, Techniques, and Datasets for Autonomous Navigation in Simulation."
- Get clearer images of the same stool or another still.
- Get a small photogrammetric model running. Hopefully we can have it in Unreal Engine soon too!
- Attend at least one event from the ones offered by HMC.
- Work through a tutorial for Unreal Engine.
- I want to go through some Unix commands tutorials