Conclusion

 

    In this project, we trained a CNN object detection model at desktop platform and and applied the trained mode into a mobile platform. As a baseline, we have a running Android app that runs our CNN model trained by TensorFlow offline.

    The accuracy is close to 100% for stationary object in the above mentioned categories in good illumination. However when one of the factors such as light of the and object from different categories is changed the accuracy is reduced.