A Framework of Transfer Learning in Object Detection for Embedded Systems
This work addresses the need for efficient object detection in real-time embedded applications, but it is incremental as it builds on existing SqueezeDet methods.
The paper tackled the problem of accelerating training for object detection on embedded systems using transfer learning with SqueezeDet, achieving a 1.8 times speedup in training through an optimized pipeline and introducing a hyperparameter optimization mechanism.
Transfer learning is one of the subjects undergoing intense study in the area of machine learning. In object recognition and object detection there are known experiments for the transferability of parameters, but not for neural networks which are suitable for object detection in real time embedded applications, such as the SqueezeDet neural network. We use transfer learning to accelerate the training of SqueezeDet to a new group of classes. Also, experiments are conducted to study the transferability and co-adaptation phenomena introduced by the transfer learning process. To accelerate training, we propose a new implementation of the SqueezeDet training which provides a faster pipeline for data processing and achieves 1.8 times speedup compared to the initial implementation. Finally, we created a mechanism for automatic hyperparameter optimization using an empirical method.