Our implementation refers to the DeepPhase[9] methodology. The methodologies proposed in the DeepPhase are for Cataract surgeries, while our project is based on Cholecystectomy surgeries. Hence, we had to make some architectural changes to best suit our problem.
We referred to an already available codebase for DeepPhase[12]. This codebase do not belong to the authors of DeepPhase, hence, we first validated the code and after making significant changes, we built our implementation on top of this modified codebase. Several new architectures were tried, their implementations were written, and added as part of the experiments in order to reach our desired results. Also, some of the other changes include a custom frame extraction mechanism, configuration changes, parameter tuning, etc. We have added a README file containing specific instructions to make the code runnable.