forresti/SqueezeNet issues and pull requests

#65 - Some minor mistakes in the paper

Issue - State: open - Opened by Alphacch about 1 year ago - 3 comments

#64 - squeezenet for speech

Issue - State: open - Opened by akankshaaa13 about 3 years ago - 3 comments

#63 - squeezenet v1_1 for facedetector , possible , feasable ?

Issue - State: open - Opened by blackholeearth over 3 years ago - 1 comment

#62 - Image normalization values

Issue - State: closed - Opened by AlexMuresan almost 5 years ago - 1 comment

#61 - optimization and compression

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

#60 - why not use lr_mult, decay_mult like {1, 1, 2, 0}?

Issue - State: open - Opened by ujsyehao about 6 years ago - 3 comments

#59 - which label list you used

Issue - State: closed - Opened by azuryl about 6 years ago - 1 comment

#58 - Fine-tuning SqueezeNet

Issue - State: closed - Opened by mun3 over 6 years ago - 2 comments

#57 - Added link to a Matlab-compatible model

Pull Request - State: open - Opened by titsitits over 6 years ago

#56 - The SqueezeNet deploy.caffemodel files have all 0.0 weight and bias data

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

#55 - Update README with community implementation link

Pull Request - State: closed - Opened by vonclites over 6 years ago - 1 comment

#54 - SqueezeNet training on cifar

Issue - State: closed - Opened by GitZinc almost 7 years ago - 3 comments

#53 - why can not get the output of the prob layer?

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

#52 - training from scratch, random seed

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

#51 - SqueezeNet is slower when using GPU than when using CPU?

Issue - State: closed - Opened by shunsuke227ono about 7 years ago - 2 comments

#50 - 1.1 deploy.prototxt

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

#49 - tensorflow- After hundreds of epochs, my total_loss stay around 0.6~0.7, and not decreased

Issue - State: closed - Opened by turboLIU about 7 years ago - 2 comments

#48 - Image Width Issue

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

#47 - Top-1 Acc=61.0% on ImageNet, without any sacrificing compared with SqueezeNet v1.1.

Issue - State: open - Opened by miaow1988 over 7 years ago - 4 comments

#46 - SqueezeNet v1.1 with Residual Connections with Dense→Sparse→Dense (DSD) Training

Issue - State: open - Opened by ShervinAr over 7 years ago

#45 - Added link to CoreML implementation

Pull Request - State: closed - Opened by mdering over 7 years ago

#44 - model convert

Issue - State: closed - Opened by wm901115nwpu over 7 years ago - 1 comment

#43 - Add 1.1 deploy.prototxt

Pull Request - State: closed - Opened by cyberfire over 7 years ago - 1 comment

#42 - Has anyone successfully trained Squeezenet with residual connections?

Issue - State: closed - Opened by stoneyang over 7 years ago

#41 - SqueezeNet with Deep Compression

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

#40 - Number of anchors reducing mAP

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

#39 - Would you alos share your caffe training/val log file as well?

Issue - State: closed - Opened by Coderx7 over 7 years ago - 1 comment

#38 - How to use this SqueezeNet question?

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

#37 - Train SqueezeNet from scratch ?

Issue - State: closed - Opened by phongnhhn92 over 7 years ago - 3 comments

#36 - Pre-trained model that DSD technique is applied?

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

#35 - Small changes for faster convergence

Pull Request - State: closed - Opened by NikolasMarkou over 7 years ago - 5 comments

#34 - SqueezeNet in PyTorch

Issue - State: closed - Opened by Maratyszcza over 7 years ago - 4 comments

#33 - v1.1 dimensions mismatch in the first layers

Issue - State: closed - Opened by gudovskiy almost 8 years ago - 1 comment

#32 - conv10 layer has pad 1

Issue - State: closed - Opened by psyhtest almost 8 years ago - 3 comments

#31 - In V1.1 train_val.prototxt, why are the TRAIN & TEST phase (in loss and accuracy layers) commented out?

Issue - State: closed - Opened by aurotripathy almost 8 years ago - 1 comment

#30 - Replace global pooling with explicitly defined window

Issue - State: closed - Opened by psyhtest almost 8 years ago - 5 comments

#29 - SqueezeNet speed slower than Alexnet

Issue - State: closed - Opened by PearlDzzz almost 8 years ago - 4 comments

#28 - why cost 2500+ MB when training but the model only 5MB big?

Issue - State: closed - Opened by aceimnorstuvwxz almost 8 years ago - 2 comments

#27 - Padding in conv-10 layer

Issue - State: closed - Opened by AlexandreBriot almost 8 years ago - 1 comment

#26 - Any one run SqueezeNet by opencv dnn module??

Issue - State: closed - Opened by chibai about 8 years ago - 4 comments

#25 - binarized SqueezeNet

Issue - State: closed - Opened by AGhiuta about 8 years ago

#24 - crop_size is 227 instead of 224?

Issue - State: closed - Opened by shesung about 8 years ago - 1 comment

#23 - Image Preprocessing for stated top5 accuracy

Issue - State: closed - Opened by fervorarc about 8 years ago - 2 comments

#22 - build my own sqeezenet

Issue - State: closed - Opened by zhaishengfu about 8 years ago - 2 comments

#21 - Create deploy.prototxt

Pull Request - State: closed - Opened by ae86208 about 8 years ago - 3 comments

#20 - add deploy file

Pull Request - State: closed - Opened by rmekdma over 8 years ago

#19 - SqueezeNet out of memory with batch size (512) smaller than AlexNet (1024)

Issue - State: closed - Opened by wenwei202 over 8 years ago - 11 comments

#18 - v1.1 loss does not decrease

Issue - State: closed - Opened by kli-casia over 8 years ago - 17 comments

#17 - squeezenet of resnet?

Issue - State: closed - Opened by davidleon over 8 years ago - 1 comment

#16 - regarding performance improvement for AlexNet

Issue - State: closed - Opened by williamjames1 over 8 years ago - 20 comments

#15 - Darknet/YOLO or Caffe/R-FCN + Squeezenet

Issue - State: closed - Opened by Grabber over 8 years ago - 18 comments

#14 - why the conv10 in SqueezeNet1.0 has a pad of 1?

Issue - State: closed - Opened by cc946079647 over 8 years ago - 1 comment

#13 - Layer for Feature Extraction

Issue - State: closed - Opened by ghost over 8 years ago - 7 comments

#12 - Why v1.1 does not have deploy.prototxt

Issue - State: closed - Opened by changyun79 over 8 years ago - 10 comments

#11 - Reproduce AlexNet results.

Issue - State: closed - Opened by mrgloom over 8 years ago - 3 comments

#10 - ReLU after classifier

Issue - State: closed - Opened by ducha-aiki over 8 years ago - 1 comment

#9 - Fine-tuning SqueezeNet

Issue - State: closed - Opened by mtngld over 8 years ago - 7 comments

#8 - About the training data resolution

Issue - State: closed - Opened by austingg over 8 years ago - 2 comments

#7 - Comparing GPU memory usage to BVLC reference CaffeNet

Issue - State: closed - Opened by gavinmh over 8 years ago - 2 comments

#6 - Wrong Numbers?

Issue - State: closed - Opened by tmatas over 8 years ago - 1 comment

#5 - Type of squeezenet_v1.0.caffemodel is PCX image (image/x-pcx)?

Issue - State: closed - Opened by saustar over 8 years ago - 3 comments

#4 - Initialization weights

Issue - State: closed - Opened by liuyipei over 8 years ago - 7 comments

#3 - SqueezeNet benchmark

Issue - State: closed - Opened by ducha-aiki over 8 years ago - 10 comments

#2 - add deploy

Pull Request - State: closed - Opened by terrychenism over 8 years ago - 4 comments

#1 - deploy.prototxt

Issue - State: closed - Opened by kaishijeng over 8 years ago - 3 comments

GitHub / forresti/SqueezeNet issues and pull requests