wujixiu / helmet-detection Goto Github PK
View Code? Open in Web Editor NEWHelmet Detection on Construction Sites
License: Apache License 2.0
Helmet Detection on Construction Sites
License: Apache License 2.0
here appear
Check failed: registry.count(type) == 1 (0 vs. 1) Unknown layer type: MaskPooling
I can not find MaskPooling in hardhat-wearing-detection\SSD-RPA\src\caffe\proto\caffe.proto
I am a fresher
想知道这个在python3.5 环境下如何运行
是复现了对比的论文中的fast rcnn么,用自己的数据集重新训练的?能不能提供一下fast rcnn 的预训练模型,我想试试
Labeled region is the head but not the helmet.
It's a trick? raise the mAP?
If labeled region includes only the helmet, would the model converge faster, or not?
I found that some images are duplicates but have different names. Some of them appear both in test and train set and some others appear either two times in the train set or two times in the test set.
The images names are the following:
./00376.jpg
./00444.jpg
./00239.jpg
./00676.jpg
./00177.jpg
./02078.jpg
./00286.jpg
./02142.jpg
./00292.jpg
./02542.jpg
./02423.jpg
./03286.jpg
./02224.jpg
./03365.jpg
./00615.jpg
./00641.jpg
./00801.jpg
./01007.jpg
./00509.jpg
./00832.jpg
./00545.jpg
./01132.jpg
./00024.jpg
./00151.jpg
./00095.jpg
./00618.jpg
./01726.jpg
./03087.jpg
Hello, I am very glad that you have made contributions to this field. May I ask where can you download your data set? It was not found in the file you uploaded.Thank you very much!
have a lot of false detection, detecting people without helmet as people.
The content is not in english and unable to download the weight files, can you please provide alternate link to directly download the files?
when i run the command line make all -j12
i found this error :
PROTOC src/caffe/proto/caffe.proto make: protoc: Command not found CXX src/caffe/layer_factory.cpp Makefile:632: recipe for target '.build_release/src/caffe/proto/caffe.pb.cc' failed make: *** [.build_release/src/caffe/proto/caffe.pb.cc] Error 127 make: *** Waiting for unfinished jobs.... CXX src/caffe/common.cpp src/caffe/common.cpp:1:10: fatal error: boost/thread.hpp: No such file or directory #include <boost/thread.hpp> ^~~~~~~~~~~~~~~~~~ compilation terminated. src/caffe/layer_factory.cpp:4:10: fatal error: boost/python.hpp: No such file or directory #include <boost/python.hpp> ^~~~~~~~~~~~~~~~~~ compilation terminated. Makefile:575: recipe for target '.build_release/src/caffe/common.o' failed make: *** [.build_release/src/caffe/common.o] Error 1 Makefile:575: recipe for target '.build_release/src/caffe/layer_factory.o' failed make: *** [.build_release/src/caffe/layer_factory.o] Error 1
你好,我的QQ是1018949282,不知道能不能加一下你的联系方式,谢谢!
I downloaded yolov3-tiny_final.weights and yolov3-tiny.cfg and compile model like this
helmet_net = cv2.dnn.readNetFromDarknet(
'cfg/yolov3-tiny.cfg', 'model-weights/yolov3-tiny_final.weights')
helmet_net.setPreferableBackend(cv2.dnn.DNN_BACKEND_OPENCV)
helmet_net.setPreferableTarget(cv2.dnn.DNN_TARGET_CPU)
blob = cv2.dnn.blobFromImage(image, 1 / 255, (416, 416),
[0, 0, 0], 1, crop=False)
helmet_net.setInput(blob)
# Runs the forward pass to get output of the output layers
outs = helmet_net.forward(get_outputs_names(helmet_net))
after this I map outs to get only items where confidences are >0.7 and get following result
What can be a problem of this behavior? Actually it didn't find real helmet...
could you please share your value of final loss when you train the models like ssd or ssd-rpa, when I train it, the loss can't descent to a small value(the min loss: about 3.xxx).
很遗憾,没有这个数据库不能下载全文,能否发一份论文全文给我,非常感谢!
[email protected]
你好,我前些天也用SSD算法识别安全帽,不过是基于tensorflow的,准确度大概70%,我想知道你是怎么制作训练样本的,不知道能不能发几个样本图片到我的邮箱?[email protected].顺便想问下你识别的准确度是多少啊?最后谢谢你分享到github上。
good job, could you share the model weight files again? It seems that download link(both google driver and baidu) is not available
Thank you for your wonderful work. I wonder if I can fine tuning the yolo3-tiny model using your code? If so, how should I do it? Thanks a lot.
Your model is great!
How many training images are enough? such as to raise the mAP to 70%.
I have only a few images(100+) so my own model's mAP is lower than 50%.
I don't know if I need to append some training data?
F1008 18:36:51.900141 31887 layer_factory.hpp:81] Check failed: registry.count(type) == 1 (0 vs. 1) Unknown layer type: MaskPooling (known types: AbsVal, Accuracy, AnnotatedData, ArgMax, BNLL, BatchNorm, BatchReindex, Bias, Concat, ContrastiveLoss, Convolution, Crop, Data, Deconvolution, DetectionEvaluate, DetectionOutput, Dropout, DummyData, ELU, Eltwise, Embed, EuclideanLoss, Exp, Filter, Flatten, HDF5Data, HDF5Output, HingeLoss, Im2col, ImageData, InfogainLoss, InnerProduct, Input, LRN, LSTM, LSTMUnit, Log, MVN, MemoryData, MultiBoxLoss, MultinomialLogisticLoss, Normalize, PReLU, Parameter, Permute, Pooling, Power, PriorBox, RNN, ReLU, Reduction, Reshape, SPP, Scale, Sigmoid, SigmoidCrossEntropyLoss, Silence, Slice, SmoothL1Loss, Softmax, SoftmaxWithLoss, Split, TanH, Threshold, Tile, VideoData, WindowData)
I built the caffe and trained Pelee-RPA for 120000 iterations. Now, I am trying to make inference and it gave me this error. Could anyone have an idea about this problem? - Thanks.
Hello @wujixiu, can you please add LICENSE file to your repository?
I would like to use some of the pretrained models (Squeezenet-SSD and SSD-RPA300) on your dataset but it is not clear what license they are available under. Currently the is no license information which technically means that your repo falls under "all rights reserved".
A declarative, efficient, and flexible JavaScript library for building user interfaces.
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google ❤️ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.