nihui
/
ncnn
Not watched
1
0
Fork 0
Code
Releases 46 Wiki evaluate Activity
Issues 0 Pull Requests 0 Datasets Model Cloudbrain HPC
You can not select more than 25 topics Topics must start with a chinese character,a letter or number, can include dashes ('-') and can be up to 35 characters long.
1855 Commits
5 Branches
354 MB
2 Download
 
 
 
 
 
 
Tree: ba70b2f780
Branches Tags
${ item.name }
Create branch ${ searchTerm }
from 'ba70b2f780'
${ noResults }
ncnn/src/layer
nihui ef3550b52f
gru and rnn layer (#2572) 		5 years ago
..
arm fix onnx2ncnn adaptive pool conversion bug, remove no used weights node (#2570) 5 years ago
mips Support negative axis in concat, slice and softmax (#2365) 5 years ago
riscv Ci rv32imc (#1940) 6 years ago
vulkan support PyTorch AdaptiveAvgPool2d and AdaptiveMaxPool2d (#2546) 5 years ago
x86 pooling_x86_unused_opt (#2565) 5 years ago
absval.cpp layer creator function is not necessary for built-in layers 6 years ago
absval.h move vulkan code to subdir, new layer interface create_pipeline and destroy_pipeline for post-loading works 7 years ago
argmax.cpp Support remove libstdc++ denpendency (#2030) 5 years ago
argmax.h implement ncnn blob/workspace allocator, fine-grained per-layer openmp threads control, fix #469 8 years ago
batchnorm.cpp layer creator function is not necessary for built-in layers 6 years ago
batchnorm.h move vulkan code to subdir, new layer interface create_pipeline and destroy_pipeline for post-loading works 7 years ago
bias.cpp layer creator function is not necessary for built-in layers 6 years ago
bias.h implement ncnn blob/workspace allocator, fine-grained per-layer openmp threads control, fix #469 8 years ago
binaryop.cpp Support remove libstdc++ denpendency (#2030) 5 years ago
binaryop.h format code style and setup restyled.io (#1840) 6 years ago
bnll.cpp layer creator function is not necessary for built-in layers 6 years ago
bnll.h implement ncnn blob/workspace allocator, fine-grained per-layer openmp threads control, fix #469 8 years ago
cast.cpp layer creator function is not necessary for built-in layers 6 years ago
cast.h cast between float32 and bfloat16 6 years ago
clip.cpp layer creator function is not necessary for built-in layers 6 years ago
clip.h format code style and setup restyled.io (#1840) 6 years ago
concat.cpp Support negative axis in concat, slice and softmax (#2365) 5 years ago
concat.h move vulkan code to subdir, new layer interface create_pipeline and destroy_pipeline for post-loading works 7 years ago
convolution.cpp fix convolution.cpp shadowed variables warning (#2449) 5 years ago
convolution.h format code style and setup restyled.io (#1840) 6 years ago
convolutiondepthwise.cpp Support remove libstdc++ denpendency (#2030) 5 years ago
convolutiondepthwise.h format code style and setup restyled.io (#1840) 6 years ago
crop.cpp crop vulkan vec and image, crop x86 pack4, more crop tests, fix crop with channel tail offset 5 years ago
crop.h less duplicated code for crop layer, slice axes starts from 0 6 years ago
deconvolution.cpp Support remove libstdc++ denpendency (#2030) 5 years ago
deconvolution.h utility wrapper for neon float32 bfloat16 conversion, deconvolution deconvolutiondepthwise arm fp16s fp16sa bf16s 5 years ago
deconvolutiondepthwise.cpp Support remove libstdc++ denpendency (#2030) 5 years ago
deconvolutiondepthwise.h utility wrapper for neon float32 bfloat16 conversion, deconvolution deconvolutiondepthwise arm fp16s fp16sa bf16s 5 years ago
deepcopy.cpp layer creator function is not necessary for built-in layers 6 years ago
deepcopy.h deepcopy layer 6 years ago
dequantize.cpp layer creator function is not necessary for built-in layers 6 years ago
dequantize.h core int8 inference, quantize and dequantize, net using flag, caffe2ncnn reads int8 scale table 8 years ago
detectionoutput.cpp Support remove libstdc++ denpendency (#2030) 5 years ago
detectionoutput.h mxnet-ssd wip ... 7 years ago
dropout.cpp layer creator function is not necessary for built-in layers 6 years ago
dropout.h move vulkan code to subdir, new layer interface create_pipeline and destroy_pipeline for post-loading works 7 years ago
eltwise.cpp eltwise for vec and image, fix #2473 5 years ago
eltwise.h format code style and setup restyled.io (#1840) 6 years ago
elu.cpp layer creator function is not necessary for built-in layers 6 years ago
elu.h implement ncnn blob/workspace allocator, fine-grained per-layer openmp threads control, fix #469 8 years ago
embed.cpp layer creator function is not necessary for built-in layers 6 years ago
embed.h implement ncnn blob/workspace allocator, fine-grained per-layer openmp threads control, fix #469 8 years ago
exp.cpp layer creator function is not necessary for built-in layers 6 years ago
exp.h implement ncnn blob/workspace allocator, fine-grained per-layer openmp threads control, fix #469 8 years ago
expanddims.cpp layer creator function is not necessary for built-in layers 6 years ago
expanddims.h convert mxnet squeeze expanddims, convert onnx squeeze unsqueeze 6 years ago
flatten.cpp layer creator function is not necessary for built-in layers 6 years ago
flatten.h move vulkan code to subdir, new layer interface create_pipeline and destroy_pipeline for post-loading works 7 years ago
gemm.cpp gemm layer 5 years ago
gemm.h gemm layer 5 years ago
groupnorm.cpp fuse onnx groupnorm 5 years ago
groupnorm.h fuse onnx groupnorm 5 years ago
gru.cpp gru and rnn layer (#2572) 5 years ago
gru.h gru and rnn layer (#2572) 5 years ago
hardsigmoid.cpp Support remove libstdc++ denpendency (#2030) 5 years ago
hardsigmoid.h Implemented hard sigmoid (#1046) 7 years ago
hardswish.cpp Support remove libstdc++ denpendency (#2030) 5 years ago
hardswish.h Implemented hard swish layer 6 years ago
innerproduct.cpp innerproduct aarch64 use gemm (#2521) 5 years ago
innerproduct.h int8 code refactoring wip, add int8 test 6 years ago
input.cpp layer creator function is not necessary for built-in layers 6 years ago
input.h adreno image shader + fp16 + fp16a (#1714) 6 years ago
instancenorm.cpp instancenorm without affine 5 years ago
instancenorm.h instancenorm without affine 5 years ago
interp.cpp fix interp on vec 5 years ago
interp.h add interp param align_corner (#2236) 5 years ago
layernorm.cpp layernorm layer 5 years ago
layernorm.h layernorm layer 5 years ago
log.cpp layer creator function is not necessary for built-in layers 6 years ago
log.h implement ncnn blob/workspace allocator, fine-grained per-layer openmp threads control, fix #469 8 years ago
lrn.cpp layer creator function is not necessary for built-in layers 6 years ago
lrn.h format code style and setup restyled.io (#1840) 6 years ago
lstm.cpp layer creator function is not necessary for built-in layers 6 years ago
lstm.h LSTM arm/x86 + fp16 innerproduct arm (#1881) 6 years ago
memorydata.cpp layer creator function is not necessary for built-in layers 6 years ago
memorydata.h implement ncnn blob/workspace allocator, fine-grained per-layer openmp threads control, fix #469 8 years ago
mish.cpp layer creator function is not necessary for built-in layers 6 years ago
mish.h Add mish layer (#1733) 6 years ago
mvn.cpp layer creator function is not necessary for built-in layers 6 years ago
mvn.h implement ncnn blob/workspace allocator, fine-grained per-layer openmp threads control, fix #469 8 years ago
noop.cpp relu arm fp16s 5 years ago
noop.h adreno image shader + fp16 + fp16a (#1714) 6 years ago
normalize.cpp Support remove libstdc++ denpendency (#2030) 5 years ago
normalize.h normalize support inplace 6 years ago
packing.cpp fix packing.cpp integer multiply may overflow (#2213) 5 years ago
packing.h fuse packing cast storage, binaryop image shader, dummy buffer and image, device-wide utility packing converter operators, fix multi-blob layer test 6 years ago
padding.cpp some shared arm bf16s fp16s implementation 5 years ago
padding.h format code style and setup restyled.io (#1840) 6 years ago
permute.cpp layer creator function is not necessary for built-in layers 6 years ago
permute.h move vulkan code to subdir, new layer interface create_pipeline and destroy_pipeline for post-loading works 7 years ago
pixelshuffle.cpp pixelshuffle nhwc mode, convert onnx DepthToSpace mode DCR, convert mlir tf.DepthToSpace 5 years ago
pixelshuffle.h pixelshuffle nhwc mode, convert onnx DepthToSpace mode DCR, convert mlir tf.DepthToSpace 5 years ago
pooling.cpp support PyTorch AdaptiveAvgPool2d and AdaptiveMaxPool2d (#2546) 5 years ago
pooling.h support PyTorch AdaptiveAvgPool2d and AdaptiveMaxPool2d (#2546) 5 years ago
power.cpp layer creator function is not necessary for built-in layers 6 years ago
power.h implement ncnn blob/workspace allocator, fine-grained per-layer openmp threads control, fix #469 8 years ago
prelu.cpp layer creator function is not necessary for built-in layers 6 years ago
prelu.h move vulkan code to subdir, new layer interface create_pipeline and destroy_pipeline for post-loading works 7 years ago
priorbox.cpp Support remove libstdc++ denpendency (#2030) 5 years ago
priorbox.h add mobilenetv3-ssd (#1335) 6 years ago
proposal.cpp Support remove libstdc++ denpendency (#2030) 5 years ago
proposal.h implement ncnn blob/workspace allocator, fine-grained per-layer openmp threads control, fix #469 8 years ago
psroipooling.cpp Support remove libstdc++ denpendency (#2030) 5 years ago
psroipooling.h implement psroipooling layer 7 years ago
quantize.cpp layer creator function is not necessary for built-in layers 6 years ago
quantize.h core int8 inference, quantize and dequantize, net using flag, caffe2ncnn reads int8 scale table 8 years ago
reduction.cpp Support remove libstdc++ denpendency (#2030) 5 years ago
reduction.h format code style and setup restyled.io (#1840) 6 years ago
relu.cpp Support remove libstdc++ denpendency (#2030) 5 years ago
relu.h move vulkan code to subdir, new layer interface create_pipeline and destroy_pipeline for post-loading works 7 years ago
reorg.cpp pixelshuffle nhwc mode, convert onnx DepthToSpace mode DCR, convert mlir tf.DepthToSpace 5 years ago
reorg.h reorg nhwc mode, code format 5 years ago
requantize.cpp layer creator function is not necessary for built-in layers 6 years ago
requantize.h format code style and setup restyled.io (#1840) 6 years ago
reshape.cpp implement full permute tag support for reshape 6 years ago
reshape.h implement full permute tag support for reshape 6 years ago
rnn.cpp gru and rnn layer (#2572) 5 years ago
rnn.h gru and rnn layer (#2572) 5 years ago
roialign.cpp Fixed compile warning due to default cast (#2201) 5 years ago
roialign.h Improve ROIAlign (accelerate ROIAlign, support sampling ratio and aligned ROIAlign) (#1820) 6 years ago
roipooling.cpp Support remove libstdc++ denpendency (#2030) 5 years ago
roipooling.h implement ncnn blob/workspace allocator, fine-grained per-layer openmp threads control, fix #469 8 years ago
scale.cpp layer creator function is not necessary for built-in layers 6 years ago
scale.h move vulkan code to subdir, new layer interface create_pipeline and destroy_pipeline for post-loading works 7 years ago
selu.cpp layer creator function is not necessary for built-in layers 6 years ago
selu.h implemented selu activation (#1051) 7 years ago
shufflechannel.cpp fix shufflechannel.cpp integer multiply may overflow (#2212) 5 years ago
shufflechannel.h shufflechannel reverse group style 6 years ago
sigmoid.cpp layer creator function is not necessary for built-in layers 6 years ago
sigmoid.h move vulkan code to subdir, new layer interface create_pipeline and destroy_pipeline for post-loading works 7 years ago
slice.cpp Support negative axis in concat, slice and softmax (#2365) 5 years ago
slice.h implement ncnn blob/workspace allocator, fine-grained per-layer openmp threads control, fix #469 8 years ago
softmax.cpp Support negative axis in concat, slice and softmax (#2365) 5 years ago
softmax.h move vulkan code to subdir, new layer interface create_pipeline and destroy_pipeline for post-loading works 7 years ago
softplus.cpp softplus inplace (#2156) 5 years ago
softplus.h softplus inplace (#2156) 5 years ago
split.cpp relu arm fp16s 5 years ago
split.h adreno image shader + fp16 + fp16a (#1714) 6 years ago
spp.cpp Support remove libstdc++ denpendency (#2030) 5 years ago
spp.h format code style and setup restyled.io (#1840) 6 years ago
squeeze.cpp layer creator function is not necessary for built-in layers 6 years ago
squeeze.h convert mxnet squeeze expanddims, convert onnx squeeze unsqueeze 6 years ago
statisticspooling.cpp layer creator function is not necessary for built-in layers 6 years ago
statisticspooling.h add statisticspooling layer (#1768) 6 years ago
swish.cpp layer creator function is not necessary for built-in layers 6 years ago
swish.h Add Swish layer (#1799) 6 years ago
tanh.cpp layer creator function is not necessary for built-in layers 6 years ago
tanh.h move vulkan code to subdir, new layer interface create_pipeline and destroy_pipeline for post-loading works 7 years ago
threshold.cpp layer creator function is not necessary for built-in layers 6 years ago
threshold.h implement ncnn blob/workspace allocator, fine-grained per-layer openmp threads control, fix #469 8 years ago
tile.cpp layer creator function is not necessary for built-in layers 6 years ago
tile.h implement ncnn blob/workspace allocator, fine-grained per-layer openmp threads control, fix #469 8 years ago
unaryop.cpp code clean 6 years ago
unaryop.h format code style and setup restyled.io (#1840) 6 years ago
yolodetectionoutput.cpp Support remove libstdc++ denpendency (#2030) 5 years ago
yolodetectionoutput.h move vulkan code to subdir, new layer interface create_pipeline and destroy_pipeline for post-loading works 7 years ago
yolov3detectionoutput.cpp Support remove libstdc++ denpendency (#2030) 5 years ago
yolov3detectionoutput.h Add yolov3detectionoutput test and AVX optimization (#1994) 5 years ago
Delete task

Are you sure you want to delete this task? Once this task is deleted, it cannot be recovered.

cancel
confirm

No Description

C++ C Python Text Protocol Buffer other