TFFRCN_Resnet50
This is the faster rcnn based on Tensorflow and resnet50
fork from the TFFRCN website is https://github.com/CharlesShang/TFFRCNN
just change the python file factory.py , __ init __.py ,networks.py in /lib/networks
add resnet50 networks
resnet101 hasn't been tested
You can find how to use it from the above website
the model file is convert from the caffemodel, you can download from the baiduyun
link: http://pan.baidu.com/s/1eSuUO1s pwd: 24cf
the fine-tune model link: http://pan.baidu.com/s/1nuUYfMh pwd:5bve
resnet101 imagenet file link http://pan.baidu.com/s/1i5odxNv pwd:d8y4
USing voc07_trainval to train the Resnet50 and test on the voc07_test
The result is : Mean AP = 0.7124
AP for aeroplane = 0.7801
AP for bicycle = 0.7931
AP for bird = 0.6836
AP for boat = 0.5750
AP for bottle = 0.4892
AP for bus = 0.8322
AP for car = 0.8412
AP for cat = 0.8380
AP for chair = 0.5186
AP for cow = 0.7550
AP for diningtable = 0.6369
AP for dog = 0.7861
AP for horse = 0.7966
AP for motorbike = 0.7692
AP for person = 0.7765
AP for pottedplant = 0.4454
AP for sheep = 0.7131
AP for sofa = 0.7074
AP for train = 0.8150
AP for tvmonitor = 0.6953