search

JasonDCox / ML-Mentorship-GovSchool

0 stars 0 forks source link

Write Conclusions #37

Closed gavinjalberghini closed 2 years ago

gavinjalberghini commented 2 years ago

Due March 7th

Acceptance Criteria:

brandonC1234 commented 2 years ago

Written Conclusion:

As the results show, tensorRT and YOLO can both provide adequate performance for computer vision on the NVIDIA Jetson Nano, however, we found that YOLO development was easier for beginners compared to Tensorflow. While tfLite does set aside resources that could be used for other tasks in an application, it’s performance and lack of GPU utilization makes it unsatisfactory unless the application specifically needs it. In all, this application of a pet doorbell system is very feasible on the NVIDIA Jetson with both Tensorflow’s tensorRT and YOLO’s Tinyv4 algorithms providing adequate speed and accuracy for the task.