Table 3 SSD learning parameters
From: An automated fruit harvesting robot by using deep learning
Architecture | Caffe |
|---|---|
Net | VGG-16 |
Image (trainval) | 200 images (1081 apples) |
Image (test) | 50 images (259 apples) |
Base learning rate | 0.0001 |
Batch size | 4 |
Learning times | 10,000 steps |