Junho Cho

Junho Cho

AI/ML Research Engineer

๐Ÿ“ Seoul
http://github.com/junhocho
๋…ผ๋ฌธ ๊ด€๋ฆฌ with Zotero + Synology Nas webDAV

๋…ผ๋ฌธ ๊ด€๋ฆฌ with Zotero + Synology Nas webDAV

Zotero ย ๋Š” ๋…ผ๋ฌธ ๊ด€๋ฆฌ ํ”„๋กœ๊ทธ๋žจ์ด๋‹ค. ๋น„์Šทํ•œ ํ”„๋กœ๊ทธ๋žจ์œผ๋กœ Mendeley, Papers ๋“ฑ์ด ์žˆ๋‹ค. ๋น„์Šทํ•œ ๋…ผ๋ฌธ(์„œ์ง€) ํ”„๋กœ๊ทธ๋žจ ์ค‘์— ๋‚ด๊ฐ€ zotero๋ฅผ ์“ฐ๋Š” ์ด์œ  ๋‹จ์ˆœํžˆ ๋ฌด๋ฃŒ์ด๊ธฐ ๋•Œ๋ฌธ์ด๋‹ค. ์˜คํ”ˆ์†Œ์Šค ํ”„๋กœ๊ทธ๋žจ์ด๊ธฐ ๋•Œ๋ฌธ์— ํฐ ๊ธฐ๋Œ€์•ˆํ•˜๊ณ  ๋Œ€ํ•™์› ๋‹ค๋‹๋™์•ˆ ์ž˜ ์‚ฌ์šฉํ•˜์˜€๋Š”๋ฐ, ์ง€๊ธˆ์€ ๋”์šฑ ์™„์„ฑ๋„ ์žˆ๋Š” ํ”„๋กœ๊ทธ๋žจ์ด ๋˜์—ˆ๋‹ค. ํƒ€ ํ”„๋กœ๊ทธ๋žจ์ฒ˜๋Ÿผ macOS, iPadOS, ์œˆ๋„์šฐ (์•ˆ๋“œ๋กœ์ด๋“œx) ๋“ฑ์„ ์ง€์›ํ•˜๋ฉฐ ์›น๋ธŒ๋ผ์šฐ์ ธ ํ™•์žฅํ”„๋กœ๊ทธ๋žจ์—์„œ ๋…ผ๋ฌธ์„ ๋ฐ”๋กœ

macOS์—์„œ ํ‚ค๋ณด๋“œ re-mapping
macOS

macOS์—์„œ ํ‚ค๋ณด๋“œ re-mapping

๋‚˜์˜ macOS ํ‚ค๋ณด๋“œ ๋ฆฌ๋งคํ•‘ ์ด์œ  esc ํ‚ค๊ฐ€ ๋„ˆ๋ฌด ๋ฉ€๋‹ค. ํ”ํžˆ ์“ฐ์ด์ง€ ์•Š๋Š” capslock์„ ๋Œ€์‹  ์ผ์œผ๋ฉฐ ์ข‹๊ฒ ๋‹ค. ํ•œ์˜ํ‚ค๊ฐ€ ์œˆ๋„์šฐ ํ‚ค๋ณด๋“œ ๊ฐ™์ด ๊ฐ™์€ ์œ„์น˜์— ์žˆ์œผ๋ฉด ์ข‹๊ฒ ์Œ. ์™ธ๋ถ€ ํ‚ค๋ณด๋“œ๋Š” right_alt ๋ฒ„ํŠผ, ๋‚ด์žฅ ํ‚ค๋ณด๋“œ๋Š” right_cmd๋ฅผ ํ•œ์˜ํ‚ค๋กœ ์‚ฌ์šฉํ•˜๊ฒ ๋‹ค. ์™ธ๋ถ€ ํ‚ค๋ณด๋“œ ์—ฐ๊ฒฐ์‹œ window ํ‚ค๋Š” cmd, altํ‚ค๋Š” optionํ‚ค๊ฐ€ ๋˜์–ด ๋ณธ๋ž˜ ์• ํ”Œํ‚ค๋ณด๋“œ์—์„œ ์“ฐ๋Š” ๊ฒŒ ๋ฐ˜๋Œ€๋กœ ๋˜๋Š”

Google HackFair ๋Œ์•„๋ณด๊ธฐ - MetaMong
Computer Vision

Google HackFair ๋Œ์•„๋ณด๊ธฐ - MetaMong

2015๋…„ ์„์‚ฌ ์ž…ํ•™ ํ›„ ์—ฐ๋ง์ฏค์— ์†Œ์†Œํ•˜๊ฒŒ ํ–ˆ๋˜ ํ”„๋กœ์ ํŠธ. ์˜ˆ์ „ ์ธํ„ฐ๋ทฐํ–ˆ๋˜๊ฒŒ ์ƒ๊ฐ๋‚˜์„œ ๊ณต์œ . ์ธํ„ฐ๋ทฐ link 0. Google HackFair๋Š” ๊ตฌ๊ธ€ ๊ธฐ์ˆ ์„ ์ด์šฉํ•ด์„œ ๋งŒ๋“  ๋‹ค์–‘ํ•œ ๊ฒฐ๊ณผ๋ฌผ๋“ค์„ ์ „์‹œํ•˜๊ณ  ๊ณต์œ ํ•˜๋Š” ํ–‰์‚ฌ์ž…๋‹ˆ๋‹ค. ์ง€๋‚œ 12์›” 5์ผ, ํ”„๋กœ์ ํŠธ ์ฐธ๊ฐ€์ž๋ถ„๋“ค์€ ์•ˆ๋“œ๋กœ์ด๋“œ, Cardboard, TensorFlow ๋“ฑ ๋‹ค์–‘ํ•œ ๊ตฌ๊ธ€ ๊ธฐ์ˆ ์„ ์‚ฌ์šฉํ•œ ์žฌ๋ฏธ์žˆ๋Š” ํ”„๋กœ์ ํŠธ๋ฅผ ์ „์‹œํ•˜๊ณ  ํ’์„ฑํ•œ ๋ณผ๊ฑฐ๋ฆฌ์™€ ๊ธฐ์ˆ ๋ ฅ ๊ทธ๋ฆฌ๊ณ  ๋ฒ”์ƒ์น˜ ์•Š์€ ์—…๋ ฅ(?)์„

BicycleGAN : Image Translation with GAN (5)
Computer Vision

BicycleGAN : Image Translation with GAN (5)

Limitations of pix2pix, DTN, DiscoGAN & CycleGAN? They produce single answer. They are deterministic models. Translates an image in one-to-one Paired set, One-to-One : pix2pix (CVPR2017) Unpaired set, One-to-One : DTN (ICLR2017), CycleGAN (ICCV2017) Paired set, One-to-Many : ??? BicycleGAN: Toward Multimodal Image-to-Image Translation (NIPS2017) BicycleGAN github Easy approach: Adopt stochastically sampled noise $N(

Ghost 1.0.0 ๋กœ ์—…๊ทธ๋ ˆ์ด๋“œ + ๋ฐ€๋ฆฐ ํฌ์ŠคํŠธ๋“ค
ghost

Ghost 1.0.0 ๋กœ ์—…๊ทธ๋ ˆ์ด๋“œ + ๋ฐ€๋ฆฐ ํฌ์ŠคํŠธ๋“ค

1 ๊ท€์ฐฎ์•„์„œ ์—…๊ทธ๋ ˆ์ด๋“œ๋ฅผ ๋ฏธ๋ฃจ๊ณ  ์žˆ์—ˆ๋Š”๋ฐ, ๊ฑฐ๋Œ€ ์—…๋ฐ์ดํŠธ๋ฅผ ํ•œ์ฐธ ์ „์— ํ–ˆ๊ธธ๋ž˜ ๊ฐ„๋งŒ์— ์—…๊ทธ๋ ˆ์ด๋“œํ–ˆ๋‹ค. ๊ฐ„๋งŒ์— ๋ธ”๋กœ๊ทธ ๋งŒ์ง€์ž‘ ๊ฑฐ๋ฆฌ๋Š”๊ฒŒ ์žฌ๋ฐŒ์—ˆ๋Š”๋ฐ ์ด๋‚ด ์—…๊ทธ๋ ˆ์ด๋“œ๋Š” ์ปค๋‹ค๋ž€ ๊ท€์ฐฎ์Œ์„ ๊ฐ€์ ธ๋‹ค์ฃผ์—ˆ๋‹ค. ๊ฒฐ๋ก ์ ์œผ๋กœ ์„œ๋ฒ„์˜ ๋ฆฌ๋ˆ…์Šค๊ฐ€ Ubuntu 16.04๊ฐ€ ์•„๋‹ˆ๋ฉด ์ž์ž˜ํ•œ ์—๋Ÿฌ๊ฐ€ ๋งŽ๋‹ค. ์ด๋ฒˆ ๋ฒ„์ „๋ถ€ํ„ฐ ghost-cli๋ฅผ ํ†ตํ•ด v1์ด์ƒ์—์„œ ์‰ฝ๊ฒŒ ์—…๊ทธ๋ ˆ์ด๋“œ๊ฐ€ ๊ฐ€๋Šฅํ† ๋ก cli๋ฅผ ์ง€์›ํ•œ๋‹ค. ์—ญ์‹œ ๊ฐ“๊ฐ“ cli๋‹ค. ๊ทธ๋ฆฌ๊ณ  ํž˜๋“ค๊ฒŒ Nginx๋“ฑ์„ ์„ค์ •ํ•ด์คฌ์—ˆ๋Š”๋ฐ(์ฒ˜์Œ

Command Line Interface
Linux

Command Line Interface

dotfiles and materials available at @junhocho[1] GUI๋ณด๋‹ค ์ข‹์€ CUI, CUI. Command Line Interface๋ฅผ ์“ฐ์‹œ๋ฉด ์–ด๋””์„œ๋“  ์‰ฝ๊ฒŒ, ์ธํ„ฐ๋„ท๋งŒ ์žˆ๊ณ , Terminal์ด๋‚˜ Putty๊ฐ€ ์žˆ์œผ๋ฉด ๊ฐœ๋ฐœ์„ ํ•  ์ˆ˜ ์žˆ๋‹ค. ์ด๋Ÿฐ ํŒ€๋ทฐ์–ด ์—†์ด๋„ ๋ง์ด๋‹ค. CLI. ๊ทธ๋ž˜์„œ ์–ด๋–ป๊ฒŒ? ( Tmux + VIM ) ๋ฅผ ์‚ฌ์šฉํ•˜์ž CLI๋ฅผ ์‚ฌ์šฉํ•˜๋Š” ์ด์œ . Portable (์ธํ„ฐ๋„ท์ด ๊ตฌ๋ฆฌ๋ฉด ํŒ€๋ทฐ์–ด๋Š” ํž˜๋“ค์ง€) Simple Automated (์›ํ•˜๋Š” ๊ธฐ๋Šฅ์€ ๋‹ค

home automation

Home Automation memo

์ค‘์š”! ์ƒค์˜ค๋ฏธ ๋ฉ€ํ‹ฐํƒญ ํŠน๊ฐ€ ์˜ˆ์‚ฐ ์กฐ๋ช… ํ•„๋ฆฝ์Šค ํœด 2 + ๋ธŒ๋ฆฟ์ง€ = 90 ์ƒค์˜ค๋ฏธ ์Šคํƒ ๋“œ = 28.1 ( 37.5 - 5.9 () - 3.5 (PAYCO) ) (Qoo10. ๊ทธ์ „์— G9๊ฐ€ 39๋กœ ์ ค์ŒŒ์Œ) ์ƒค์˜ค๋ฏธ ์ „๊ตฌ RGBW 26 ์ค‘๊ณ ๋‚˜๋ผ ๋Œ€๋ฆฌ๊ตฌ๋งค (32.6 / ์ค‘๊ณ ๋‚˜๋ผ 20) ์ƒค์˜ค๋ฏธ ์ „๊ตฌ W 17 + 10(๋ฐฐ์†ก๋น„) ์Œ์„ฑ์ธ์‹ ์—์ฝ”๋‹ท2 = 70 ๊ตฌ๊ธ€ํ™ˆ = 160

TensorFlow-v1.0.0 + Keras ์„ค์น˜ (Windows/Linux/macOS)
Deep Learning

TensorFlow-v1.0.0 + Keras ์„ค์น˜ (Windows/Linux/macOS)

์ฐธ๊ณ  :https://groups.google.com/forum/#!topic/keras-users/_hXfBOjXow8 ์„ ์š”์•ฝ: # export PATH=~/anaconda/bin:$PATH # MAC conda create -n tf python=3.5 # 17/3/1 ๊ธฐ์ค€์œผ๋กœ ์œˆ๋„์šฐ์—์„œ 3.5 ๋ฒ„์ „๋งŒ TensorFlow/Keras๊ฐ€ ์ง€์› activate tf # Windows # source activate tf : Linux/macOS # ์—ฌ๊ธฐ์„œ๋ถ€ํ„ฐ (tf) ํ™˜๊ฒฝ. ์„ค์น˜ ์ˆœ์„œ ์ค‘์š” pip install

ghost

Ghost์—์„œ AWS S3 Storage๋ฅผ ์ด๋ฏธ์ง€ ์„œ๋ฒ„๋กœ ์‚ฌ์šฉํ•˜๊ธฐ

๋ณดํ†ต Ghost์—์„œ ๊ธ€ ์“ฐ๋‹ค๊ฐ€ ์ด๋ฏธ์ง€๋ฅผ ์ฒจ๋ถ€ ํ•˜๊ณ ์‹ถ์–ด ![]()์„ ํƒ€์ดํ•‘ํ•˜๋ฉด ์š”๋Ÿฐ ์ด์œ ์—…๋กœ๋“œ ์ฐฝ์ด previewํŽ˜์ด์ง€์— ๋– ์„œ Drag-and-Drop ๊ฐ™์€ ๋ฐฉ์‹์œผ๋กœ ์‰ฝ๊ฒŒ ์—…๋กœ๋“œ ํ•  ์ˆ˜ ์žˆ๋‹ค. ํ•˜์ง€๋งŒ ์ด๋ฏธ์ง€๋Š” ๋ธ”๋กœ๊ทธ๊ฐ€ ๋Œ์•„๊ฐ€๊ณ  ์žˆ๋Š” ๋กœ์ปฌ ์„œ๋ฒ„์•ˆ์— ์ €์žฅ์ด ๋œ๋‹ค. ๊ทธ๋Ÿฐ๋ฐ ๊ทธ๋Ÿฐ ๋ฐฉ์‹์—์„œ ๋ช‡๊ฐ€์ง€ ๋‹จ์ ์ด ๋ณด์ด๋ฉด์„œ ์ด๋ฏธ์ง€๋Š” ๋‹ค๋ฅธ ์„œ๋ฒ„๋ฅผ ๋‘์–ด์„œ ์ €์žฅํ•ด๋‘๊ณ  ์‹ถ์—ˆ๋‹ค. ์›๋ž˜๋Š” Dropbox์— ์ €์žฅํ•˜๋ฉด์„œ public link๋ฅผ ๋งค๋ฒˆ

Web-scraping with Chrome extension
dev

Web-scraping with Chrome extension

Crawling, Scraping ๋น„์Šทํ•œ ๋œป์ด๋ฉฐ ๋‚ด๊ฐ€ ์›ํ•˜๋Š” ๋ฐ์ดํ„ฐ๋ฅผ ์›น์—์„œ ์ˆ˜์ง‘ํ•จ์„ ์˜๋ฏธํ•œ๋‹ค. Design Seeds ๋ผ๋Š” ์›น์‚ฌ์ดํŠธ๋ฅผ ์—ฐ๊ตฌ์˜ ๋ชฉ์ ์œผ๋กœ ํฌ๋กค๋ง์„ ํ–ˆ๋‹ค. ์ด ์›น์‚ฌ์ดํŠธ๋Š” ์ƒ‰๊ฐ์—์„œ ์˜๊ฐ์„ ๋ฐ›์•„, ์•„๋ฆ„๋‹ค์šด ์ด๋ฏธ์ง€์—์„œ ์ฃผ์š” ์ƒ‰์ƒ์„ color palette๋กœ ๋‚˜ํƒ€๋‚ด์ค€๋‹ค. ์‹ค์ œ ํ•ด๊ฒฐํ•œ ๋ฐฉ๋ฒ•์€ ํ•ด๊ฒฐ ์ด๋ž€ ์•„๋ž˜ ํ—ค๋”๋ฅผ ์ฐธ๊ณ  ์ด ์›น์‚ฌ์ดํŠธ๋ฅผ ํฌ๋กค๋งํ•˜๊ธฐ ์œ„ํ•ด ๊ฐ€์žฅ ๋จผ์ € ์ƒ๊ฐํ•œ ๋ฐฉ๋ฒ•์€ ์ด ์›น์‚ฌ์ดํŠธ์˜ Instagram ๊ณ„์ •

Code ๊ณต์œ ํ•  ๋•Œ Gist ์‚ฌ์šฉํ•ด๋ณด๊ธฐ
dev

Code ๊ณต์œ ํ•  ๋•Œ Gist ์‚ฌ์šฉํ•ด๋ณด๊ธฐ

Blog ๋˜๋Š” code ๊ณต์œ ์— ์žˆ์–ด์„œ Syntax highlight๋Š” ๊ฐ€๋…์„ฑ๋•Œ๋ฌธ์— ์ค‘์š”ํ•˜๋‹ค๊ณ  ์ƒ๊ฐํ•œ๋‹ค. ํ˜„์žฌ ์‚ฌ์šฉํ•˜๊ณ  ์žˆ๋Š” Ghost์˜ MarkDown์„ ์‚ฌ์šฉํ•˜์—ฌ code๋ฅผ ๊ณต์œ ํ•˜๋ฉด ์•„๋ž˜์™€ ๊ฐ™๊ณ . #!/bin/bash PATH="$PATH:/usr/bin/" export USER="junho" #Current unbuntu id DISPLAY="3" #์ด๊ฒƒ๋„ ์ƒˆ๋กœ๋งŒ๋“ค์–ด์คŒ. DEPTH="16" GEOMETRY="1280x800"

Torch Wiki
Deep Learning

Torch Wiki

๊ฐœ์ธ์ ์œผ๋กœ Torch๋ฅผ ์ฃผ๋ ฅ์œผ๋กœ ์‚ฌ์šฉํ•˜๋ คํ•ฉ๋‹ˆ๋‹ค. ํ•„์š”ํ•  ๋•Œ๋งˆ๋‹ค ๋น ๋ฅด๊ฒŒ ๊ฒ€์ƒ‰ํ•˜๊ธฐ ์œ„ํ•ด ๋ชจ์•„๋†“์€ ๊ฒƒ๋“ค์„ wiki ์ฒ˜๋Ÿผ ๊ณต์œ ํ•ฉ๋‹ˆ๋‹ค. Official Homepage Installation is easy. Tutorials Very useful 15 min Lua Torch Video Tutorials Bayarea DL summmer slide ref when needed demos DOCS tutorial Useful Libraries Torchnet torchnet is a framework for torch which provides a

LSTM ๊ณผ ResNet
Deep Learning

LSTM ๊ณผ ResNet

์„œ๋ก  ์šฐ์„  LSTM์€ ์ตœ๊ทผ์— ๋‚˜์˜จ ๊ฐœ๋…์ด ์•„๋‹™๋‹ˆ๋‹ค. ์ด๋ฏธ 1997๋…„ [Hochreiter et al., 1997]์—์„œ ๋‚˜์˜จ ๊ฐœ๋…์ด๋ฉฐ ๊ธฐ์กด RNN(Vanilla Recurrent Neural Network)๊ฐ€ ์˜ค๋ž˜์ „ ์ •๋ณด๋ฅผ ์žŠ์–ด๋ฒ„๋ฆฌ๋Š” ๋‹จ์ ์„ ๋ณด์™„ํ–ˆ์Šต๋‹ˆ๋‹ค. LSTM์˜ ๊ธฐ๋ณธ ๋ฐฐ๊ฒฝ์ง€์‹์„ ๋‹ค ์„ค๋ช…ํ•˜๋ฉด ์ข‹๊ฒ ์ง€๋งŒ, ์ด ๊ธ€์€ ๊ธฐ๋ณธ์ ์œผ๋กœ LSTM์ด ์–ด๋–ป๊ฒŒ ์ž‘๋™ํ•˜๋Š”์ง€ ResNet์€ ๋ฌด์—‡์ธ์ง€ ์˜ˆ์ „์— ๊ณต๋ถ€ํ•˜๋˜๊ฑธ ์˜ฎ๊ฒจ๋…ผ ๊ฒƒ์ž…๋‹ˆ๋‹ค. ์ด ๊ธ€์€ ์ €์˜ ์ˆœ์ˆ˜ํ•œ

VNC๋กœ ์„œ๋ฒ„๋ฅผ gui๋กœ ๋‹ค๋ฃจ๊ธฐ
VNC

VNC๋กœ ์„œ๋ฒ„๋ฅผ gui๋กœ ๋‹ค๋ฃจ๊ธฐ

๋ณดํ†ต ์„œ๋ฒ„์—์„œ ์ž‘์—…์„ ํ•  ๋•Œ ํ„ฐ๋ฏธ๋„ ssh๋งŒ ์‚ฌ์šฉํ•˜์ง€๋งŒ, ๊ฐ€๋” gui๊ฐ€ ํ•„์š”ํ•  ๋•Œ๊ฐ€ ์žˆ๋‹ค. ์˜ˆ๋ฅผ ๋“ค์–ด, ๋ธŒ๋ผ์šฐ์ €๋ฅผ ์ผœ์„œ ๋กœ๊ทธ์ธ์„ ํ•˜์—ฌ ์ธ์ฆ์„ ๋ฐ›์•„์•ผ ๋‹ค์šด๋ฐ›์„ ์ˆ˜ ์žˆ๋Š” ๋งํฌ๊ฐ€ ์žˆ์„๋•Œ. ๊ทธ๋ž˜์„œ VNC๋ฅผ ์‚ฌ์šฉํ•˜์—ฌ ์„œ๋ฒ„์˜ gui๋ฅผ ์‚ฌ์šฉํ•ด๋ณด๊ณ ์ž ํ•œ๋‹ค. HOST server๋Š” UBUNTU 14.04, client๋Š” macOS๋ผ๊ณ  ๊ฐ€์ •ํ•˜์˜€์ง€๋งŒ ๋‹ค๋ฅธ ๊ฒฝ์šฐ๋„ ๋ณ„๋ฐ˜ ๋‹ค๋ฅด์ง€ ์•Š์„ ๊ฒƒ์ด๋‹ค. ์ฃผ์š” ์•„์ด๋””์–ด๋Š” VNC-server๋ฅผ

adobe

๋งฅ(macOS)์—์„œ Adobe ์ œํ’ˆ๋“ค์˜ ์ˆ˜๋™ ์—…๋ฐ์ดํŠธ

OSX๊ฐ€ macOS๋กœ Sierra์œผ๋กœ ๋ฐ”๋€Œ๋ฉด์„œ ๋‚ก์€ ์ •๋ณด์ผ ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค. Illustrator 2014CC๋กœ ์ž‘์—…์„ ํ•˜๋Š”๋ฐ, ๋”๋ธ”/ํŠธ๋ฆฌํ”Œ ๋ชจ๋‹ˆํ„ฐ ๋“ฑ์—์„œ ์šฐํด๋ฆญ์ด ์•ˆ๋˜๋Š” ๋ฒ„๊ทธ๊ฐ€ ์žˆ๋‹ค. ์ด๋ฅผ ํ•ด๊ฒฐํ•˜๊ธฐ ์œ„ํ•ด์„œ๋Š” ํŒจ์น˜๋ฅผ ํ•ด์•ผํ•œ๋‹ค๊ณ ํ•œ๋‹ค. ๊ทผ๋ฐ Creative App ์ž๋™์—…๋ฐ์ดํŠธ ์–ด์ฉŒ๊ตฌ์ €์ฉŒ๊ตฌ๋Š”ํ•˜๋ฉด ๋˜๋Š”๋ฐ, ๋‚ด๊ฐ€ ๋ฐ›์€ Illustrators๋Š” ์„œ์šธ๋Œ€ SW๋ผ ๋ถˆ๊ฐ€๋Šฅํ•œ๋Œ„๋‹ค. ์ด๊ฒƒ์„ ํ•ด๊ฒฐํ•˜๊ธฐ ์œ„ํ•ด์„œ ์ˆ˜๋™์—…๋ฐ์ดํŠธ๋ฅผ ๊ฒ€์ƒ‰์œผ๋กœ [์ด๊ณณ](์ด๊ณณ์—์„œ ์—์„œ ๋ฐ›์•˜๋‹ค. ๊ทธ๋Ÿผ์—๋„ ๋ถˆ๊ตฌํ•˜๊ณ  ๊ด€๋ฆฌ์ž ์–ด์ฉŒ๊ตฌ

Weakly supervised Learning์œผ๋กœ Object localizationํ•˜๊ธฐ
Detection

Weakly supervised Learning์œผ๋กœ Object localizationํ•˜๊ธฐ

๊ณต์œ ํ•˜๊ณ ์žํ•˜๋Š” ๋…ผ๋ฌธ์˜ ์ œ๋ชฉ์€ Learning Deep Features for Discriminative Localization Bolei Zhou, Aditya Khosla, Agata Lapedriza, Aude Oliva, Antonio Torralba Computer Science and Artificial Intelligence Laboratory, MIT ๋…ผ๋ฌธ์˜ ๋ฐ๋ชจ๋Š” "์—ฌ๊ธฐ"์„œ ์‰ฝ๊ฒŒ ํ•ด๋ณผ ์ˆ˜ ์žˆ๋‹ค. ๋ฐ๋ชจ์˜ ๋‚ด์šฉ์„ ํ’€์ดํ•˜์ž๋ฉด ๋จผ์ € ์ฒ˜์Œ์—” scene label๋กœ ํƒœ๊น…๋˜์–ด ์žˆ๋Š” ๋ฐ์ดํ„ฐ๋กœ Weakly supervised learning์„ ํ•œ