The aim of this dashboard is to provide to the user a cool, simple to use interface to compare deep learning models according to multiple performance criterias such as: Accurcay, Latency, Throughput, Size and Memory consumption.
Now, the comparaison is made based on pre-collected measures. In the future, we aim to provide a real-time comparaison on the hardware platform that the program is running on, allowing for easy multi-hardware comparaison. The user can just download the code on any hardware plateform and run it to get a full comparaison of the models on the platform. the only condition is that the platform contains a browser and python installed.
The dashboard was created with Plotly, an interactive graphing library for Python. The Interface is modified from the Fifa dashboard project