Gossip of A.I. — #024

莊鐵頭
BlendVision
Published in
2 min readDec 18, 2017

NIPS Top 10 Institutions

  • Google + DeepMind : 13.4%
  • Industrial : 19.3%, Academic: 33.1%
  • Where is Facebook & OpenAI?
  • 業界:19.3%,學界:33.1%
  • 沒有 FacebookOpenAI

Landing.ai

Defect detection examples.

Looks like a consulting company to me. The video shows examples that can be improved by A.I.: important, simple but exhausting works.

看起來像顧問公司。影片中有可用機器學習解決的實例問題:重要、簡單但是費時費力。

Microsoft Extends AirSim to Include Autonomous Car Research

An environment for autonomous car research. OpenAI had a similar environment but has been removed (deepdrive / OpenAI universe) a year ago.

微軟 AirSim 提供自駕車研究環境。OpenAI 曾經釋出過類似的環境,但是因為法律問題移除了 (deepdrive / OpenAI universe)。

TopoSketch

We can interpolate faces if we have their latent vectors. Check the demo to see a new proposed method to create animations.

如果知道臉部圖片的特徵,就可以內插臉的表情。TopoSketch 用連結特徵向量的方式來建立動畫。

Light Cascaded Convolutional Neural Networks for Accurate Player Detection

Use convolutional neural networks to locate players’ on the court. There are some use case here:

  • During a live broadcast, reporter may need the positions to show possible strategies.
  • Sometimes people may want to build animations of a game and it’s difficult to label all positions (over entire game).

使用卷積網路定位球場上的選手。實際需求:

  • 賽事轉播中球評可能需要球員的位置來講解戰術。
  • 把比賽的錄影以動畫方式呈現(視角可能不同,標記所有選手在整場比賽的位置費時費力)

High-Resolution Image Synthesis and Semantic Manipulation with Conditional GANs

Pix2Pix is HD now! Check the video to see the future of photo creator. (I was thinking about the “creator”. Should I use creator or editor? It’s possible to edit pictures base on the pix2pix technic. It might work by translating a picture to its labeled version, editing the labeled image and translating it back to picture domain. Cycle consistence is required.)

Pix2Pix 的高解析度版本,影片中有修圖的應用。直接編輯相片是可行的(目前的技術效果不會好),首先將相片轉換成色塊標籤版本,修改標籤再轉換回相片的版本。

ironhead replicated pix2pix

ironhead replicated cyclegan

--

--