Official Tensorflow Implementation of Deep Metric Learning with Chance Constraints (Chance Constrained Programming for Deep Metric Learning)
Implements many dml methods and provides a benchmarking framework.
A detailed guide will be prepared soon.
tensorflow >= 2.8
yaml, numpy, PIL, sklearn, scipy, matplotlib, imageio, pprint
(1) put custom config files in ./metric_learning/configs
(2) run >python trainModel.py command with arguments
(2.1) dataset can be passes as an argument
(2.2) dataset is downloaded automatically
(3) different configurations can be experimented by changing related .yaml files