Decide on Comparable Dataset and Test/Train Split

JasonDCox / ML-Mentorship-GovSchool

0 stars 0 forks source link

Closed JasonDCox closed 2 years ago

JasonDCox commented 2 years ago

Decide on a dataset to train and implement onto the Jetsons as well as the train/test split. Decide on metrics to test by.

Find a way to have the same pictures for test-train split in formats for both YOLO and Tensorflow
Get metric output from algorithms with common dataset. ex. runtime, accuracy.

brandonC1234 commented 2 years ago

For the general dog breeds, we will be using the by-species version of the Oxford pet dataset, found here

For the specific dog, we will share identical pictures (found in the dataset folder of the GitHub) and annotate them for the algorithms.

For this project, we will focus on FPS, GPU usage, CPU usage, memory usage, average precision, and mean average precision

GPU usage, CPU usage, and memory usage will be gathered using tegra-stats