Computer Vision News - March 2017

Now comes the real part, training and testing the R-CNN network: 14 Computer Vision News Tool Tool Preparing Download network parameters final-0000.params from: https://github.com/dmlc/mxnet-model-gallery and the model files themselves here: http://data.dmlc.ml/models/ Get datasets wget http://data.dmlc.ml/mxnet/models/imagenet/vgg/vgg16-0000.params wget http://data.dmlc.ml/mxnet/models/imagenet/resnet/101-layers/resnet- 101-0000.params the result is selective_search_data wget http://www.cs.berkeley.edu/~rbg/fast-rcnn-data/selective_search_data.tgz Testing usage: test.py [-h] [--image_set IMAGE_SET] [--year YEAR] [--root_path ROOT_PATH] [--gpu GPU_ID] Test a Fast R-CNN network optional arguments: -h, --help show this help message and exit --image_set can be test --year can be 2007, 2010, 2012 --root_path output data folder For example: python test.py --network resnet --image_set 2007_trainval+2012_trainval --gpu 1 Training end2end on the VOC / COCO dataset Train_end2end.py with the following parameters --network - could be resnet , vggnet, alexnet --image_set could be 2007_trainval or something like 2007trainval+2012trainval. --dataset_path could be something like data/VOCdevkit, where images, annotations and results can be put For example: python train_end2end.py --network resnet --image_set 2007_trainval+2012_trainval --gpu 1 python test.py --network resnet --gpu 1 Training on your own model See more info here: https://github.com/deboc/py-faster-rcnn/tree/master/help Resources: Tianqi Chen, Mu Li, Yutian Li, Min Lin, Naiyan Wang, Minjie Wang, Tianjun Xiao, Bing Xu, Chiyuan Zhang, and Zheng Zhang - MXNet: A Flexible and Efficient Machine Learning Library for Heterogeneous Distributed Systems. In Neural Information Processing Systems, Workshop on Machine Learning Systems, 2015