README.md

MobileNet-SSD

A caffe implementation of MobileNet-SSD detection network, with pretrained weights on VOC0712 and mAP=0.727.

Run

1. Download SSD source code and compile (follow the SSD README).
2. Download the pretrained deploy weights from the link above.
3. Put all the files in SSD_HOME/examples/
4. Run demo.py to show the detection result.
5. You can run merge_bn.py to generate a no bn model, it will be much faster.

Train your own dataset

1. Convert your own dataset to lmdb database (follow the SSD README), and create symlinks to current directory.

ln -s PATH_TO_YOUR_TRAIN_LMDB trainval_lmdb
ln -s PATH_TO_YOUR_TEST_LMDB test_lmdb

1. Create the labelmap.prototxt file and put it into current directory.
2. Use gen_model.sh to generate your own training prototxt.
3. Download the training weights from the link above, and run train.sh, after about 30000 iterations, the loss should be 1.5 - 2.5.
4. Run test.sh to evaluate the result.
5. Run merge_bn.py to generate your own no-bn caffemodel if necessary.

python merge_bn.py --model example/MobileNetSSD_deploy.prototxt --weights snapshot/mobilenet_iter_xxxxxx.caffemodel

About some details

There are 2 primary differences between this model and MobileNet-SSD on tensorflow:

1. ReLU6 layer is replaced by ReLU.
2. For the conv11_mbox_prior layer, the anchors is [(0.2, 1.0), (0.2, 2.0), (0.2, 0.5)] vs tensorflow's [(0.1, 1.0), (0.2, 2.0), (0.2, 0.5)].

Reproduce the result

I trained this model from a MobileNet classifier(caffemodel and prototxt) converted from tensorflow. I first trained the model on MS-COCO and then fine-tuned on VOC0712. Without MS-COCO pretraining, it can only get mAP=0.68.

Mobile Platform

You can run it on Android with my another project rscnn.