py-RFCN-priv
py-RFCN-priv copied to clipboard
soeaver
code for py-R-FCN-multiGPU maintained by bupt-priv
``` [ 2%] Building NVCC (Device) object src/caffe/CMakeFiles/cuda_compile_1.dir/layers/cuda_compile_1_generated_batch_reindex_layer.cu.o /usr/local/cuda/include/thrust/system/cuda/detail/cub/device/../iterator/arg_index_input_iterator.cuh(120): error: KeyValuePair is not a template /usr/local/cuda/include/thrust/system/cuda/detail/cub/device/dispatch/../../agent/agent_reduce.cuh(65): error: identifier "BlockReduceAlgorithm" is undefined /usr/local/cuda/include/thrust/system/cuda/detail/cub/device/dispatch/../../agent/agent_reduce.cuh(66): error: identifier "CacheLoadModifier" is undefined /usr/local/cuda/include/thrust/system/cuda/detail/cub/device/dispatch/../../agent/agent_reduce.cuh(76): error: identifier...
``` def bbox_transform_inv(boxes, deltas): if boxes.shape[0] == 0: return np.zeros((0, deltas.shape[1]), dtype=deltas.dtype) boxes = boxes.astype(deltas.dtype, copy=False) widths = boxes[:, 2] - boxes[:, 0] + 1.0 heights = boxes[:, 3] -...
could you please share a demo for fpn, and describe how you built fpn? thank you very much @soeaver
~~~ py-RFCN-priv/caffe-priv/src/caffe/layers/box_annotator_ohem_layer.cu(50): warning: lambda expressions are a C++11 feature py-RFCN-priv/caffe-priv/src/caffe/layers/box_annotator_ohem_layer.cu(49): error: a template argument may not reference a local type detected during instantiation of "void caffe::BoxAnnotatorOHEMLayer::Forward_gpu( const std::vector &, const...
我编译没出现问题，但是要训练模型的时候出现这个问题，是不是caffe版本的问题？ 训练脚本是这样的： python ./tools/train_net.py\ --gpu 0 \ --solver models/pascal_voc/ResNet-101/rfcn_end2end/solver_ohem.prototxt \ --weights data/imagenet_models/ResNet-101-model.caffemodel \ --imdb voc_2007_trainval \ --iters 110000 --cfg experiments/cfgs/rfcn_end2end_ohem_pascal_voc.yml \ --rand
https://github.com/hszhao/PSPNet/blob/master/evaluation/prototxt/pspnet101_VOC2012_473.prototxt#L30
Deformable psroipooling code can be found in this repo, so is deformable convolution also implemented?
