Fine tune avg-vgg16 on CUB200 dataset

Remove fc layers, replace pool5 with 14x14 average pooling and softmax layer, then fine tune on CUB200.

How to use

• modify generate_prototxt.py to specify the path of caffe, lmdb
• ./fine_tune.sh

Implementation details

• lmdb: 256*256 cub200 dataset(#train: 5994, #val:5794).
• adopt random cropping to generate 224*224 images when training and use center cropping when test.
• training issues:
  • learning rate of softmax is 10x larger;
  • batch size:32, 188 iters/epoch, train 21 epoch;
  • use 2 gpu cards;
  • learning rate change 10^-3~10^-5 (decay 10x every 7 epoch);

Results

epoch	val acc	epoch	val acc	epoch	val acc
1	0.4294	2	0.5299	3	0.5614
4	0.5740	5	0.6051	6	0.5564
7	0.5732	8	0.7083	9	0.7130
10	0.7154	11	0.7106	12	0.7125
13	0.7126	14	0.7140	15	0.7145
16	0.7149	17	0.7147	18	0.7145
19	0.7132	20	0.7119	21	0.7118

Others

This strategy is much better than train softmax first follwed with fine tuning, which only achieving 0.70176 accuracy!