Phase 1 Specification This phase was to specify the major project objective and requirements to set the major core of the project to guide the later stage. Major researches were done at this phase to set guidelines for later phases. Major software choices were made in this stage to confirm the constraints.
Phase 2 Software and hardware testing This phase was to assure that the choices from phase 1 was feasible. The major work done was to learn to use and produce tests for justification. However this phase cannot test the performance because the real data for the final output was still not yet prepared.
- Summary of work done
- All required component including hardware and software installed or purchased
- All to be linked component tested compatible. For example, TensorFlow installed on Raspberry Pi.
- Observable fundamental functions worked. For example, the arm can move according to signals transported from the program codes designed for it, simple classification program can run with Keras
Phase 3 Construction This phase is to construct a final model working model based on the deliverable from earlier stages. This phase has gone through 5 major training trial sets of experiments to create a satisfying final model. Requirements and justification of trial sets were made according to previous phases and were updated according to findings of each experiment. Details and evaluation of the experiment will be reported in the final year project report
Phase 4 Modification and wrap up This phase improved the deliverable of the previous phase and wrapped up the whole project. Since project time is limited, an ending point of the project was set at this stage, and work done in this stage was to wrap up the whole project to prepare for presentation and exhibition.
- Work done in this stage:
- Expanded and refined data set (data set trial 5B-D) for enhancing the output of the project deliverable of phase 3, to be explained in the final year porjest
- Implemented YOLO algorithm for world coordinate extraction of items to be recognized