Comments (6)
There are many examples showing how to use a pre-trained model to do object detection in MXNet, such as:
You can get started by following their README files and substitute the base CNN with DPN.
from dpns.
@Harold-Zhang Thank you for your interest.
In the competition, we used the Faster R-CNN framework and substituted the base CNN with different kinds of DPNs and did the ensembling. Several tricks are used to boost the performance and you can check our team's abstract at [here] for more information.
Besides the workshop [slides], we also have a [paper] and a [video](in chinese) talking about the proposed DPNs.
As for the training time, DPN-92 takes about one day using a cluster with 40 x K80 cards. The actual training speed of each node is shown at [here]. I haven't tried to train DPNs on a single node, but I guess it may take you more than a week to train DPN-92 from scratch.
from dpns.
from dpns.
@cypw Thanks for your reply. I saw the result of ILSVRC2017 Object detection on Image-Net, and your team used the ensemble of DPN models to detect objects, how did it work? Did you use DPNs? Or Faster R-CNN? Thank you~
from dpns.
@cypw Could you provide me some paper about your DPNs? I want to learn more but I didn't find more paper except ILSVRC2017 workshop. By the way, could you tell me how long it took you to train DPN-92 model with 4 x K80? Thanks for your help.
from dpns.
Thank you very much!@cypw
from dpns.
Related Issues (20)
- caffe model? HOT 5
- Why "num_filter=(num_1x1_c+2*inc)" instead of "num_filter=(num_1x1_c+inc)" HOT 2
- "Mean-Max pooling" HOT 1
- Cann't find pretrained model based on Keras
- DPN-98 on Place365-Standard dataset HOT 2
- .rec must use the DPNs's Mxnet im2rec? HOT 12
- How to replace augmenter in new mxnet source by image_aug_torch.cc
- Train Log about DPNs
- ImageNet-5k data HOT 4
- Finetune on new dataset HOT 4
- Need substrated mean RGB = [124,117,104] on DPN-68 of imagenet-5k? HOT 4
- The order of Blocks: BN-Act-Conv2d or Conv2d-BN-Act?
- some question about hyper parameter
- In DPN86-5k and DPN92-5k, What's the relation between label and semantic class ? HOT 2
- OSError HOT 1
- How to train DPNs HOT 2
- about inference image shape HOT 3
- Segmentation fault HOT 1
- How to trian my dataset by DPNs? HOT 2
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 dpns.