Profile

Cover photo
iceflame co
Works at ZJU
Lives in Hangzhou, Zhejiang, China
29 followers|23,865 views
AboutPostsCollectionsVideos

Stream

iceflame co

Shared publicly  - 
 
haahahahahahaaha
 
Google robots declare war on Christmas
You know Dasher and Dancer and Prancer and Vixen,You know Comet and Cupid and Donner and Blitzen,But do you recall,Google's most dangerous reindeer of all? Spot, the Boston Dynamics' robot,Has a...
15 comments on original post
2
Add a comment...

iceflame co

Shared publicly  - 
 
Improving  video thumbnails with DNN
2
Add a comment...

iceflame co

Shared publicly  - 
 
 
Code, models, and dataset are available at:
http://personal.ie.cuhk.edu.hk/~ys014/projects/Faceness/Faceness.html
In this paper, we propose a novel deep convolutional network (DCN) that achieves outstanding performance on FDDB, PASCAL Face, and AFW. Specifically, our method achieves a high recall rate of 90.99% on the challenging FDDB benchmark, outperforming the state-of-the-art method by a large margin of 2.91%. Importantly, we consider finding faces from a new perspective through scoring facial parts responses by their spatial structure and arrangement. T...
View original post
1
Add a comment...

iceflame co

Shared publicly  - 
 
 
We had been quietly working on Mandarin Chinese speech recognition since February, and just unveiled the system this week. Because Chinese is a tonal language---the same sound pronounced with different inflections has dramatically different meanings---Chinese is hard for non-native speakers (who often don't even hear the tones) to master. Many researchers therefore asked if we had to do a lot of work to adapt our English speech system to Chinese. With Deep Speech, we found end-to-end learning was able to figure out the tones by itself. As usual, the formula for success was simply SCALE: Lots of data, and a huge network. The ease of adapting Deep Speech to Chinese also makes me optimistic about new languages. In this article, Awni Hannun and Bryan Catanzaro share lots of details. Go to: https://medium.com/s-c-a-l-e/how-baidu-mastered-mandarin-with-deep-learning-and-lots-of-data-1d94032564a5
On Aug. 8 at the International Neural Network Society conference on big data in San Francisco, Baidu senior research eng…
5 comments on original post
1
Add a comment...

iceflame co

Shared publicly  - 
 
 
I'm enjoying this new G+ layout. Here's another Tall Photo I hadn't shared before. Taken in Sydney, Australia in January.  This dramatic sunset was going off and I was scrambling trying to find the best spot to take a photo. Fun times!  

Tomorrow I'm headed to South Korea for two nights. Does anyone have any suggestions on cool buildings or landscapes to photograph for my two quick nights there in Seoul? 
242 comments on original post
2
Add a comment...

iceflame co

Shared publicly  - 
 
 
Sun Worshiper -- San Francisco, CA
87 comments on original post
1
Add a comment...

iceflame co

Shared publicly  - 
 
In this tutorial, you'll uncover my complete guide to building an image search engine (CBIR system) using Python and OpenCV from start to finish.
1
Add a comment...

iceflame co

Shared publicly  - 
 
 
Accepted to ICCV 2015 (oral). We solve MRF by proposing a Convolutional Neural Network (CNN), namely Deep Parsing Network (DPN), which enables deterministic end-to-end computation in a single forward pass.
View original post
1
Add a comment...

iceflame co

Shared publicly  - 
 
 
Deep learning has shown remarkable success on some of the world’s most difficult computer science challenges. Today on the Google Research blog we go into how Long Short-term Memory Recurrent Neural Networks (LSTM RNNs) are being used to improve Google Voice transcription. 
10 comments on original post
1
Add a comment...

iceflame co

commented on a video on YouTube.
Shared publicly  - 
 
fantastic

iceflame co

commented on a video on YouTube.
Shared publicly  - 
 
no zuo no die
 ·  Translate

iceflame co

Shared publicly  - 
 
cool
 
MIT's Cheetah robot is capable of a 22 km/h pace while (theoretically) carrying a battery pack. When the robot apocalypse happens, you should be wearing your running shoes. http://www.engadget.com/2013/05/17/mit-cheetah-robot-runs-faster-more-efficiently/
85 comments on original post
1
Add a comment...
iceflame's Collections
Collections iceflame is following
Work
Employment
  • ZJU
    present
Basic Information
Gender
Male
Story
Tagline
for the beauty
Places
Map of the places this user has livedMap of the places this user has livedMap of the places this user has lived
Currently
Hangzhou, Zhejiang, China
Links