文件列表:

LICENSE (1067, 2018-10-16)
demo.py (1975, 2018-10-16)
deploy.prototxt (44667, 2018-10-16)
gen.py (17879, 2018-10-16)
gen_model.sh (1298, 2018-10-16)
images (0, 2018-10-16)
images\000001.jpg (78771, 2018-10-16)
images\000067.jpg (50854, 2018-10-16)
images\000456.jpg (105302, 2018-10-16)
images\000542.jpg (115536, 2018-10-16)
images\001150.jpg (88635, 2018-10-16)
images\001763.jpg (73424, 2018-10-16)
images\004545.jpg (123072, 2018-10-16)
merge_bn.py (4426, 2018-10-16)
mobilenet_iter_73000.caffemodel (23306119, 2018-10-16)
solver_test.prototxt (477, 2018-10-16)
solver_train.prototxt (501, 2018-10-16)
template (0, 2018-10-16)
template\MobileNetSSD_deploy_template.prototxt (44697, 2018-10-16)
template\MobileNetSSD_test_template.prototxt (45453, 2018-10-16)
template\MobileNetSSD_train_template.prototxt (47176, 2018-10-16)
test.sh (235, 2018-10-16)
train.prototxt (47136, 2018-10-16)
train.sh (349, 2018-10-16)
train_voc.sh (143, 2018-10-16)
voc (0, 2018-10-16)
voc\MobileNetSSD_deploy.prototxt (29353, 2018-10-16)
voc\MobileNetSSD_test.prototxt (45407, 2018-10-16)
voc\MobileNetSSD_train.prototxt (47149, 2018-10-16)
voc\solver.prototxt (493, 2018-10-16)

# MobileNet-SSD A caffe implementation of MobileNet-SSD detection network, with pretrained weights on VOC0712 and mAP=0.727. Network|mAP|Download|Download :---:|:---:|:---:|:---: MobileNet-SSD|72.7|[train](https://drive.google.com/open?id=0B3gersZ2cHIxVFI1Rjd5aDgwOG8)|[deploy](https://drive.google.com/open?id=0B3gersZ2cHIxRm5PMWRoTkdHdHc) ### Run 1. Download [SSD](https://github.com/weiliu89/caffe/tree/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 ``` 2. Create the labelmap.prototxt file and put it into current directory. 3. Use gen_model.sh to generate your own training prototxt. 4. Download the training weights from the link above, and run train.sh, after about 30000 iterations, the loss should be 1.5 - 2.5. 5. Run test.sh to evaluate the result. 6. 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](https://github.com/tensorflow/models/blob/master/object_detection/g3doc/detection_model_zoo.md): 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](https://drive.google.com/open?id=0B3gersZ2cHIxZi13UWF0OXBsZzA) and [prototxt](https://drive.google.com/open?id=0B3gersZ2cHIxWGEzbG5nSXpNQzA)) converted from [tensorflow](http://download.tensorflow.org/models/mobilenet_v1_1.0_224_2017_06_14.tar.gz). 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](https://github.com/chuanqi305/rscnn).

近期下载者：

相关文件：

评论：[我要评论] [举报此文件]

收藏者：