# Context For an undergraduate project at the University of Michigan, my team collected this data to see if we could classify gender using images of people. Then to try to improve performance, we used machine learning and computer vision methods to track the person over time, automatically and used the additional tracking data to boost the performance of our convolutional neural network. [It turned out that tracking improved our performance on our gender classifier.][1] # Content 70 sequences of people walking with tight bounding boxes and gender labels. CURRENT STATUS Can't upload data due, since the site detects the contents of my .zip files as uncompressed. [1]: https://csbuja.github.io/documents/EECS%20445%20Poster.pdf