Car detection using transfer learning methodology
Due to the emergence and rise of smartphones and other forms of embedded computing in today's times, it has become important to find object detection algorithms that are less resource intensive. In this project, I have tried to analyse some of the reasons existing algorithms for object detec...
Saved in:
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Final Year Project |
| Language: | English |
| Published: |
2017
|
| Subjects: | |
| Online Access: | http://hdl.handle.net/10356/70465 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Institution: | Nanyang Technological University |
| Language: | English |
| Summary: | Due to the emergence and rise of smartphones and other forms of embedded
computing in today's times, it has become important to find object detection
algorithms that are less resource intensive. In this project, I have tried to analyse
some of the reasons existing algorithms for object detection are GPU intensive
and have tried to implement an algorithm for car localization using neural
networks that is also GPU efficient. Due to limited availability of GPU resources,
it was not feasible to train the model from scratch. Hence, we have used transfer
learning techniques by reusing some of the weights of some known models which
are used for similar tasks. |
|---|