shicai/MobileNet-Caffe issues and pull requests

#91 - Wrong predictions after fine-tuning the mobilenetv2 model

Issue - State: closed - Opened by neerajasjawali about 1 year ago - 4 comments

#90 - Mobilenet-NNIE

Issue - State: open - Opened by CVUsers about 2 years ago - 2 comments

#89 - TypeError: slice indices must be integers or None or have an index method

Issue - State: open - Opened by nicolast0604 over 2 years ago - 4 comments

#88 - modified line35 and line38 type(off)=int

Pull Request - State: open - Opened by Jeremy-J-J over 3 years ago

#87 - batchnorm layer variance exist negative numbers

Issue - State: open - Opened by KimmiShi over 3 years ago - 3 comments

#86 - can you share the training code?

Issue - State: open - Opened by junhyukso almost 4 years ago

#85 - Training imagenet from scratch

Issue - State: open - Opened by makiit over 4 years ago

#84 - check failure stack trace

Issue - State: open - Opened by Endeavour92 about 5 years ago

#83 - caffe可以实现宽度乘法器和分辨率乘法器吗？

Issue - State: open - Opened by 17759205390 about 5 years ago

#82 - bug in eval_image,py

Issue - State: open - Opened by vicwer about 5 years ago

#81 - can you provide the prototxt generator ? thank you very much

Issue - State: open - Opened by 13354236170 over 5 years ago

#80 - MobileNet025

Issue - State: open - Opened by yja1 over 5 years ago

#79 - Some bottlenecks do not contain skip connections in Mobilenet v2?

Issue - State: closed - Opened by eezywu over 5 years ago

#78 - Update README.md

Pull Request - State: closed - Opened by mn-robot over 5 years ago

#77 - Check failed: target_blobs.size() == source_layer->blobs().size() (1 vs. 0) Incompatible number of blobs for layer conv5_5/dw

Issue - State: closed - Opened by ngunauj almost 6 years ago - 1 comment

#76 - The accuracy i test is lower than that in the readme file

Issue - State: open - Opened by basaltzhang almost 6 years ago - 1 comment

#75 - Fine-tune with provided solver leads to loss=87.3365

Issue - State: open - Opened by joechen24 almost 6 years ago - 6 comments

#74 - Add Netron links

Pull Request - State: closed - Opened by lutzroeder about 6 years ago

#73 - image size

Issue - State: closed - Opened by wang5566 about 6 years ago

#72 - add train.prototxt

Pull Request - State: open - Opened by ZhangXinNan about 6 years ago

#71 - I wonder something intresting why it's

Issue - State: closed - Opened by xufeifeiWHU about 6 years ago

#70 - Is it recommended to use global batch for fine tuning?

Issue - State: open - Opened by wyasuda over 6 years ago

#69 - mobilenet_v2 MAdds 429M vs paper's 300M, because of two difference

Issue - State: open - Opened by iceflame89 over 6 years ago - 1 comment

#68 - About Top1 and Top5 result

Issue - State: closed - Opened by fanghuaqi over 6 years ago - 5 comments

#67 - Wrong strides in MobilenetV2

Issue - State: closed - Opened by mathmanu over 6 years ago - 2 comments

#66 - BatchNorm consuming a lot of time

Issue - State: open - Opened by leochli over 6 years ago - 2 comments

#65 - 训练时将Conv+group换成DepthwiseConvolution更快些

Issue - State: open - Opened by jnulzl over 6 years ago - 3 comments

#64 - Loss = -nan

Issue - State: open - Opened by 3ntr0phy over 6 years ago - 3 comments

#63 - Train my own data

Issue - State: open - Opened by mauione1 over 6 years ago - 5 comments

#62 - Finetuning example

Issue - State: open - Opened by 3ntr0phy over 6 years ago

#61 - Is there any pretrained model detect human?

Issue - State: closed - Opened by Wisgon over 6 years ago - 1 comment

#60 - Where is depthwise convolution layer?

Issue - State: closed - Opened by 408550969 over 6 years ago - 4 comments

#59 - Training does not converge

Issue - State: open - Opened by SmartMachineBay over 6 years ago - 3 comments

#58 - This is incorrect implement because you use regular convolution2D rather than depthwise convolution2D

Issue - State: closed - Opened by machanic over 6 years ago - 2 comments

#57 - Bechmarking using caffe test tool

Issue - State: closed - Opened by yeshwanthv5 over 6 years ago - 4 comments

#56 - The regularization of depthwise convolution

Issue - State: open - Opened by BOBrown over 6 years ago - 1 comment

#55 - why set "BatchNorm" lr_mult: 0 ?

Issue - State: closed - Opened by xialuxi over 6 years ago - 1 comment

#54 - how to set the parameters of the model?

Issue - State: open - Opened by chenxiang204 over 6 years ago - 1 comment

#53 - MobileNetv2 SSD?

Issue - State: open - Opened by didulle over 6 years ago

#52 - MobileNet v2 CPU inferencing performance

Issue - State: open - Opened by matt-ny over 6 years ago - 3 comments

#51 - MobileNetV2 doesn't have Relu6?

Issue - State: open - Opened by wang1ang over 6 years ago - 1 comment

#50 - problem in conv4_7 layer

Issue - State: closed - Opened by linchaozhang over 6 years ago - 5 comments

#49 - MobileNet accuracy

Issue - State: open - Opened by gudovskiy almost 7 years ago - 4 comments

#48 - mobilenet_v2_deploy.prototxt is encoded as utf8-bom

Issue - State: open - Opened by sanallen almost 7 years ago - 1 comment

#47 - Unable to run eval_image.py

Issue - State: open - Opened by kushagraagrawal almost 7 years ago - 5 comments

#46 - Simply running "classification" fails... whats wrong??

Issue - State: closed - Opened by rmbarbosa almost 7 years ago - 5 comments

#45 - error == cudaSuccess (77 vs. 0 ) an illegal memory access was encountered

Issue - State: open - Opened by Ai-is-light almost 7 years ago - 2 comments

#44 - how to change the two super-parameters ?

Issue - State: closed - Opened by Ai-is-light almost 7 years ago - 3 comments

#43 - check failed::error == cudasuccess(2 vs 0) out of memory

Issue - State: open - Opened by lcj820 almost 7 years ago - 1 comment

#42 - Check failed: status == CUDNN_STATUS_SUCCESS (4 vs. 0)

Issue - State: closed - Opened by lcj820 almost 7 years ago - 11 comments

#41 - About training time for each 10 iterations

Issue - State: open - Opened by XiongweiWu almost 7 years ago - 4 comments

#40 - Can not load model weights

Issue - State: closed - Opened by BalintMagyar4 almost 7 years ago - 1 comment

#39 - How to set width multiplier parameter(alpha in paper) in caffe?

Issue - State: closed - Opened by fk1174 almost 7 years ago - 1 comment

#38 - how to test the shufflenet's classification model?

Issue - State: open - Opened by xiaocmxiao almost 7 years ago - 11 comments

#37 - Output normalization/scale?

Issue - State: closed - Opened by sipiseppala almost 7 years ago - 2 comments

#36 - what is the alpha value of the caffe-mobilenet

Issue - State: closed - Opened by samsun639 almost 7 years ago - 2 comments

#35 - Scale layer after BatchNorm

Issue - State: closed - Opened by chaosmail about 7 years ago - 7 comments

#34 - ask for help

Issue - State: closed - Opened by 22wei22 about 7 years ago - 1 comment

#33 - Require Static URL of caffemodel

Issue - State: closed - Opened by Surmeh about 7 years ago - 1 comment

#32 - why it is so slow to use the caffe engine

Issue - State: closed - Opened by andeyeluguo about 7 years ago - 5 comments

#31 - why forward processing time is larger than alexnet?

Issue - State: open - Opened by mollyStark about 7 years ago - 1 comment

#30 - So to train a Caffe for mobileNet, we need to have a custom convolutional layer right? How did you achieve training without the layer changes?

Issue - State: closed - Opened by 25b3nk about 7 years ago

#29 - Could you share your detailed training settings ?

Issue - State: open - Opened by PkuRainBow about 7 years ago - 1 comment

#28 - can not load pretrained model with the deploy prototxt

Issue - State: open - Opened by ouceduxzk about 7 years ago - 6 comments

#27 - keras model to caffe

Issue - State: open - Opened by xweihithub over 7 years ago - 1 comment

#26 - caffe

Issue - State: open - Opened by ghost over 7 years ago - 2 comments

#25 - Getting Check failed: status == CUDNN_STATUS_SUCCESS (1 vs. 0) CUDNN_STATUS_NOT_INITIALIZED

Issue - State: open - Opened by Coderx7 over 7 years ago - 3 comments

#24 - Input image preprocessing

Issue - State: closed - Opened by zvadaszi over 7 years ago - 6 comments

#23 - Not getting a good acc

Issue - State: open - Opened by leochli over 7 years ago - 6 comments

#22 - What is the mobilenet.caffemodel deploy.prototxt output ? There no output?

Issue - State: closed - Opened by siminghua over 7 years ago - 3 comments

#21 - caffemodel

Issue - State: open - Opened by wtniao over 7 years ago - 11 comments

#20 - Check failed: error == cudaSuccess (74 vs. 0) misaligned address

Issue - State: open - Opened by wendalegood over 7 years ago - 3 comments

#19 - About transform_param

Issue - State: closed - Opened by ryusaeba over 7 years ago - 2 comments

#18 - train prototxt

Issue - State: open - Opened by elliskui over 7 years ago - 3 comments

#17 - why scale: 0.017?

Issue - State: open - Opened by epleone over 7 years ago - 13 comments

#16 - about detection

Issue - State: closed - Opened by ryusaeba over 7 years ago - 1 comment

#15 - Fail to run on GPU

Issue - State: open - Opened by samsun639 over 7 years ago - 3 comments

#14 - Are top1 and top5 accuracy calculated in ILSVRC2012 val dataset?

Issue - State: closed - Opened by KeyKy over 7 years ago - 1 comment

#13 - caffe implementation speed

Issue - State: closed - Opened by GragonShit over 7 years ago - 18 comments

#12 - out of memory on 12g memory card

Issue - State: closed - Opened by NHZlX over 7 years ago - 5 comments

#11 - unclear caffe version

Issue - State: closed - Opened by davidduo over 7 years ago - 1 comment

#10 - have you tried with NVCaffe-0.16

Issue - State: open - Opened by CFAndy over 7 years ago - 4 comments

#9 - Can you post the solver?

Issue - State: closed - Opened by CFAndy over 7 years ago - 5 comments

#8 - traning time

Issue - State: closed - Opened by yunren over 7 years ago - 2 comments

#7 - training error when using engine: CAFFE

Issue - State: closed - Opened by zjchuyp over 7 years ago - 3 comments

#6 - train_val prototxt

Issue - State: closed - Opened by ShervinAr over 7 years ago - 11 comments

#5 - Where is the label file?

Issue - State: closed - Opened by chuanqi305 over 7 years ago - 1 comment

#4 - Check failed: status == CUDNN_STATUS_SUCCESS (3 vs. 0) CUDNN_STATUS_BAD_PARAM

Issue - State: closed - Opened by wlgcqh over 7 years ago - 1 comment

GitHub / shicai/MobileNet-Caffe issues and pull requests