Comments (26)
19 -> 10 -> 5 -> 3 -> 2 -> 1
from mobilenet-ssd.
I trained this on pascal VOC 2007, and I found it's mAP is only 0.6, not very good. Now I'm training it on voc 07+12 to improve it's accuracy, please wait.
from mobilenet-ssd.
@chuanqi305 I know, I am slso do this job, can I add your qq for more details?
from mobilenet-ssd.
@chuanqi305 , your model is very fast. But I have a question, your model used 5 feature maps from pre-nets, and the official SSD used 6. Is it this lead to the mAP is only 0.6?
from mobilenet-ssd.
I have tested 6 feature maps , but the mAP is virtually not changed. Google has released MobileNet SSD on tensorflow, but I can't convert it to caffemodel because there is a "ReLU6" layer. I don't want to modify the caffe source. So, now I'm training a new model like Google's but without ReLU6.
from mobilenet-ssd.
Hi, where can we get MobileNetSSD on tensorflow? I only found google api stuff. I don't get any detail about it. Thanks :)
from mobilenet-ssd.
@chuanqi305
“I have tested 6 feature maps , but the mAP is virtually not changed. Google has released MobileNet SSD on tensorflow, but I can't convert it to caffemodel because there is a "ReLU6" layer. I don't want to modify the caffe source. So, now I'm training a new model like Google's but without ReLU6.”
What's your mean about "Google has released MobileNet SSD on tensorflow, but I can't convert it to caffemodel because there is a "ReLU6" layer."? Do you mean Google's implementation including of a special activation Layer that type is "ReLU6" but not "ReLU"? As we all know, the type of "ReLU" is a basic operation for all the general networks, which can be converted to caffemodel from tensorflow easily, I think. And I wonder if you can give the link of "ReLU6" about MobileNet SSD on tensorflow?
Looking forward to further discussion!
from mobilenet-ssd.
@ryusaeba
MobileNetSSD on tensorflow is here
from mobilenet-ssd.
@csyking
Yes, ReLU6 is the new activation layer on tensorflow like this:
y = min(max(0, x), 6)
I found ReLU6 is a little better than ReLU for MobileNet.
Now I'm trying to reproduce the same MobileNet-SSD as tensorflow without ReLU6. I just upload the MobileNet classifier and the new MobileNet-SSD training prototxt very like tensorflow and the training is still in process.
If you have successfully trained the new model, please share with me.
from mobilenet-ssd.
Thanks for the reply. As my understanding, this is just API. We can't modify the network structure, right? Or I miss something?
from mobilenet-ssd.
BTW: I am also using CAFFE to do MobileNet-SSD training. Please also share with me if anyone have successfully trained the network.
from mobilenet-ssd.
@chuanqi305
Now I only train the Caffe-MobileNet-SSD on VOC 07+VOC12, and the VOC2007 test mAP≈0.65. If you have got a higher score?
I am also training it on ImageNet 200 classes detection, but the performance is worse now : (
from mobilenet-ssd.
@csyking My training result is the same as yours, mAP≈0.65.
Maybe the ReLU6 layer can improve the performance?
from mobilenet-ssd.
@ryusaeba Now you can use the gen.py to modify the network structure.
from mobilenet-ssd.
@chuanqi305 @csyking I only get mAP ~= 0.62 so I guess I need to check more detail. Per my understanding, MobileNet+SSD300 should get similar mAP with VGG+SSD300 (0.75) since Google claim this number in the following paper (https://arxiv.org/abs/1611.10012). I also attach below for your reference.
Finally,we compare against the new MobileNet network [14], which has been shown to achieve VGG-16 level accuracy on Imagenet with only 1=30 of the computational cost and model size.
Wish we can get Google's training parameters.
from mobilenet-ssd.
@chuanqi305, I have a question about your detection network. In VGG+SSD network, the first extraction is from feature map 38x38 and the second one is from feature map 19x19. But according to your gen.py, the first one and second one feature map are not similar to that in VGG+SSD. Could you please point out this?
@csyking , do you have the same question?
from mobilenet-ssd.
@ryusaeba I have the same question about feature map resolution. I just followed Google's network structure on tensorflow.
from mobilenet-ssd.
@chuanqi305, ok, in this way, how large feature map size in last extraction layer (conv17_2) is NxN?
from mobilenet-ssd.
@chuanqi305 Do you get higher mAP? So far, I still can't get similar results as paper describes.
from mobilenet-ssd.
I got mAP about 0.68, and I didn't find the VOC detection result in the paper, how much is it?
from mobilenet-ssd.
In https://arxiv.org/abs/1611.10012, they said "Finally, we compare against the new MobileNet network [14], which has been shown to achieve VGG-16 level accuracy on Imagenet with only 1=30 of the computational cost and model size." and VGG-16+SSD github said they achieve 0.77 mAP so I assume MobileNet+SSD should have similar mAP as VGG-16+SSD.
from mobilenet-ssd.
The mAP of COCO test result in the mobilenet paper is also lower than the SSD paper, so maybe the performance of mobilenet is not so good on detection task.
from mobilenet-ssd.
Ok, I get you :) Thanks for your discussion
from mobilenet-ssd.
@chuanqi305
Thanks a lot for your work.
I have a question. How can you improve the mAP, some tricks for training? because i only get about 63% mAP on VOC07.
from mobilenet-ssd.
Train the model on MS-COCO and fine tune it on VOC will make it better.
from mobilenet-ssd.
data_param {
source: "/home/chl/chl/MobileNet-SSD/caffe/examples/VOC0712/trainval_lmdb/"
batch_size: 24
backend: LMDB
error:F1017 15:19:42.540328 19268 db_lmdb.hpp:15] Check failed: mdb_status == 0 (2 vs. 0) No such file or directory
why?
@chuanqi305 @Jmx5200 @tuq820
from mobilenet-ssd.
Related Issues (20)
- ./train.sh results got aborted (core dumped) HOT 3
- Working code, but no detections using C++?
- Getting error [Errno 2] No such file or directory when I run create_data.sh HOT 3
- Error caused by 'line' and 'size' lines in create_list.sh
- Custom data training using VGG16 deploy.prototxt and .caffemodel
- 如果你是看[手把手教你]有问题进来的,赶紧把环境搞成一样的
- how to run train.sh scipt with cpu only mode
- Training aborts when running train.sh for unknown reason HOT 1
- How to evaluate results after running train.sh??
- Incompatible number of blobs for layer conv0/bn
- How to set the size of prior box of PriorBox layer?
- Accuracy too low for custom dataset
- Which data is in validation set in trainval.txt?
- 关于mobilenet中卷积核的深度的问题
- Reproduce deploy.prototxt
- # HOT 1
- how to compute meanAp for deploy.prototxt?
- where is scripts/annotate.py
- Hii HOT 1
- Object
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from mobilenet-ssd.