[B! GAN] proto_jpã®ãƒ–ãƒƒã‚¯ãƒžãƒ¼ã‚¯

proto_jp id:proto_jp

GANã«é–¢ã™ã‚‹proto_jpã®ãƒ–ãƒƒã‚¯ãƒžãƒ¼ã‚¯ (7)

${{author_name}}$

{{{comment_expanded}}}

{{label}}

{{#is_bookmark}}ãƒªã‚¹ãƒˆ{{/is_bookmark}}{{^is_bookmark}}ãƒªãƒ³ã‚¯{{/is_bookmark}}

${{author_name}}$
{{author_name}}{{created}}
{{ #comment }}{{ comment }}{{ /comment }}
- {{ label }}

${{author_name}}$

{{{comment_expanded}}}

{{label}}

{{#is_bookmark}}ãƒªã‚¹ãƒˆ{{/is_bookmark}}{{^is_bookmark}}ãƒªãƒ³ã‚¯{{/is_bookmark}}

ç”»åƒæ‹¡å¤§ãªã‚‰kakudaiAC
æ±šã„ç”»åƒã¯ã€ã‚¤ãƒ¡ãƒ¼ã‚¸ãŒæ‚ªããªã‚Šã¾ã™ã®ã§ã”æ³¨æ„ãã ã•ã„ã€‚ ç”»åƒã‚’ãã®ã¾ã¾æ‹¡å¤§ã™ã‚‹ã¨ã€ã»ã¨ã‚“ã©ã®å ´åˆã€ç”»åƒãŒæ±šããªã£ã¦ã—ã¾ã„ã¾ã™ã€‚æ±šã„ç”»åƒã¯ã€ã‚ãªãŸã®ã‚¤ãƒ¡ãƒ¼ã‚¸ã‚’æãã¦ã—ã¾ã†ã“ã¨ã«ç¹‹ãŒã‚Šã‹ãã¾ã›ã‚“ã€‚æœ€æ–°ã®äººå·¥çŸ¥èƒ½ æŠ€è¡“ã‚’é§†ä½¿ã—ãŸkakudaiACã‚’ä½¿ãˆã°ã€ç¾Žã—ã„ç”»è³ªã®ã¾ã¾ç”»åƒã‚’æ‹¡å¤§ã™ã‚‹ã“ã¨ãŒã§ãã¾ã™ã€‚ æ‹¡å¤§ã‚µã‚¤ã‚ºã¯ã€2å€ãƒ»4å€ãƒ»8å€ãƒ»16å€ã®ä¸ã‹ã‚‰è‡ªç”±ã«ãŠé¸ã³ã„ãŸã ã‘ã¾ã™ã€‚ãƒŽã‚¤ã‚ºé™¤åŽ»ã‚„ã‚·ãƒ£ãƒ¼ãƒ—æ©Ÿèƒ½ã‚‚è¨å®šãŒå¯èƒ½ã§ã™ã€‚ ESRGANã‚’ç”¨ã„ãŸè¶…è§£åƒAIã«ã‚ˆã‚Šã€é«˜è§£åƒåº¦ã®ç”»åƒã‚’ç”Ÿæˆã—ã¾ã™ã€‚ å¾“æ¥ã®ç”»åƒç·¨é›†ãƒ„ãƒ¼ãƒ«ã‚’ä½¿ã£ãŸç”»åƒã®æ‹¡å¤§ã§ã¯ã€ã¼ã‚„ã‘ãŸã‚ŠãƒŽã‚¤ã‚ºãŒå…¥ã£ãŸã‚Šã™ã‚‹ãªã©ã€å“è³ªãŒä½Žä¸‹ã—ã¾ã™ã€‚ã“ã‚Œã‚’ä¿®æ£ã™ã‚‹ã«ã¯å¤šãã®æ™‚é–“ã¨äººçš„ãƒªã‚½ãƒ¼ã‚¹ã‚’å¿…è¦ã¨ã—ã¾ã™ã€‚ æ‹¡å¤§ACã§ã¯ã€ESRGANã¨ã„ã†ãƒ‡ã‚£ãƒ¼ãƒ—ãƒ©ãƒ¼ãƒ‹ãƒ³ã‚°ã‚’ä½¿ã£ãŸAIãƒ¢ãƒ‡ãƒ«ã«ã‚ˆã‚Šã€å†™çœŸã®ç´°ã‹ãªéƒ¨åˆ†ã‚’è‡ªå‹•ã§ç”Ÿæˆã—ã€å“è³ªã‚’æãªã†ã“ã¨ãªãã€æœ€å¤§16å€ã«ã¾ã§å†™çœŸã‚’æ‹¡å¤§ã—ã¾ã™
proto_jp 2023/05/07
ai

GAN

technology

webã‚µãƒ¼ãƒ“ã‚¹
ãƒªãƒ³ã‚¯
GANã¨ã©ã†é•ã†ï¼Ÿå›ã¯ç”»åƒç”Ÿæˆãƒ¢ãƒ‡ãƒ«NeRFã‚’çŸ¥ã£ã¦ã„ã‚‹ã‹ï½¡
3ã¤ã®è¦ç‚¹ âœ”ï¸ NeRFã¨ã¯æ–°è¦è¦–ç‚¹ã®ç”»åƒç”Ÿæˆãƒãƒƒãƒˆãƒ¯ãƒ¼ã‚¯ã§ã‚ã‚‹ï½¡ âœ”ï¸Â NeRFã®å…¥åŠ›ã¯ï½¤5æ¬¡å…ƒï¼ˆç©ºé–“åº§æ¨™ã®x,y,zã¨è¦–ç‚¹ã®Î¸,Ï†ï¼‰ã§ï½¤å‡ºåŠ›ã¯ä½“ç©å¯†åº¦ï¼ˆâ‰’é€æ˜Žæ„Ÿï¼‰ã¨æ”¾å°„è¼åº¦ï¼ˆâ‰’RGBã‚«ãƒ©ãƒ¼ï¼‰ã§ã‚ã‚‹ï½¡ âœ”ï¸Â NeRFã«ã‚ˆã£ã¦å¾“æ¥ã‚ˆã‚Šã‚‚è¤‡é›‘ãªå½¢çŠ¶ã‚’æŒã¤å¯¾è±¡ç‰©ã®æ–°è¦è¦–ç‚¹ç”»åƒã‚’å¾—ã‚‹ã“ã¨ã«æˆåŠŸã—ãŸï½¡ NeRF: Representing Scenes as Neural Radiance Fields for View Synthesis writtenÂ byÂ Ben Mildenhall,Â Pratul P. Srinivasan,Â Matthew Tancik,Â Jonathan T. Barron,Â Ravi Ramamoorthi,Â Ren Ng (Submitted on 19 Mar 2020 (v1), last revised 3 Aug 2020 (this version,
proto_jp 2022/09/30
GAN

ç”»åƒ

ai

3d

algorithm
ãƒªãƒ³ã‚¯
GAN(æ•µå¯¾çš„ç”Ÿæˆãƒãƒƒãƒˆãƒ¯ãƒ¼ã‚¯)ã«ã¤ã„ã¦èª¬æ˜Žã—ã¾ã™!
GAN(æ•µå¯¾çš„ç”Ÿæˆãƒãƒƒãƒˆãƒ¯ãƒ¼ã‚¯)ã«ã¤ã„ã¦èª¬æ˜Žã—ã¾ã™! çš†ã•ã‚“ã€ã€ŒAIã€ã¨ã„ã†å˜èªžã‚’è€³ã«ã™ã‚‹ã“ã¨ãŒæœ€è¿‘å¢—ãˆã¦ããŸã®ã§ã¯ãªã„ã‹ã¨æ€ã„ã¾ã™ã€‚ã—ã‹ã—ã€ãã®ä»•çµ„ã¿ã«ã¤ã„ã¦çŸ¥ã‚ŠãŸã„ã¨æ„Ÿã˜ã¦ã‚‚ã€ã€Œç§ã«ã¯é›£ã—ãã†...ã€ã¨æ•¬é ã—ã¦ã—ã¾ã£ã¦ã„ã‚‹æ–¹ã‚‚å¤šã„ã®ã§ã¯ãªã„ã§ã—ã‚‡ã†ã‹? ã“ã“ã§ã¯ã€AIã®ãƒãƒƒãƒˆãƒ¯ãƒ¼ã‚¯ã§ã‚ã‚‹GANã«ã¤ã„ã¦ã€èª¬æ˜Žã—ã¦ã„ããŸã„ã¨æ€ã„ã¾ã™! ç›®æ¬¡ GANã¨ã¯ GANã§ä½•ãŒå‡ºæ¥ã‚‹ã®ã‹ GANã®ç¨®é¡ž å¼Šç¤¾ã®ã‚µãƒ³ãƒ—ãƒ« ã¾ã¨ã‚ GANã¨ã¯ GANã¨ã¯ã€Generative Adversarial Networkï¼ˆæ•µå¯¾çš„ç”Ÿæˆãƒãƒƒãƒˆãƒ¯ãƒ¼ã‚¯ï¼‰ã¨å‘¼ã°ã‚Œã‚‹AIã®ä¸€ç¨®ã§ã™ã€‚ GANã¯ãƒ‡ãƒ¼ã‚¿ã‹ã‚‰ç‰¹å¾´ã‚’å¦ç¿’ã™ã‚‹ã“ã¨ã§ã€å®Ÿåœ¨ã—ãªã„ãƒ‡ãƒ¼ã‚¿ã‚’ç”Ÿæˆã—ãŸã‚Šã€å˜åœ¨ã™ã‚‹ãƒ‡ãƒ¼ã‚¿ã®ç‰¹å¾´ã«æ²¿ã£ã¦å¤‰æ›ã§ãã¾ã™ã€‚ GANã®ãƒãƒƒãƒˆãƒ¯ãƒ¼ã‚¯æ§‹é€ ã¯ã€Generatorï¼ˆç”Ÿæˆãƒãƒƒãƒˆãƒ¯ãƒ¼ã‚¯ï¼‰ã¨Discriminatorï¼ˆè˜åˆ¥ãƒãƒƒãƒˆãƒ¯ãƒ¼ã‚¯ï¼‰ã®ï¼’ã¤ã®ãƒãƒƒãƒˆãƒ¯
proto_jp 2022/08/12
GAN

ai

Midjourney
ãƒªãƒ³ã‚¯
ã€è«–æ–‡ãƒ¡ãƒ¢:StackGANã€‘StackGAN: Text to Photo-realistic Image Synthesis with Stacked Generative Adversarial Networks - Re:ã‚¼ãƒã‹ã‚‰å§‹ã‚ã‚‹MLç”Ÿæ´»
è«–æ–‡ è‘—è€… èƒŒæ™¯ ç›®çš„ã¨ã‚¢ãƒ—ãƒãƒ¼ãƒ ç›®çš„ ã‚¢ãƒ—ãƒãƒ¼ãƒ ææ¡ˆæ‰‹æ³• Stage-â… GAN ãã®ä»– è£œè¶³ Stage-â…¡ GAN å®Ÿè£…ã®ãƒ‘ãƒ©ãƒ¡ãƒ¼ã‚¿ ã‚¢ãƒƒãƒ—ã‚µãƒ³ãƒ—ãƒªãƒ³ã‚°ãƒ–ãƒãƒƒã‚¯ ãƒ€ã‚¦ãƒ³ã‚µãƒ³ãƒ—ãƒªãƒ³ã‚°ãƒ–ãƒãƒƒã‚¯ ãƒ‘ãƒ©ãƒ¡ãƒ¼ã‚¿ä»– è©•ä¾¡ ãƒ‡ãƒ¼ã‚¿ã‚»ãƒƒãƒˆ æ¯”è¼ƒå¯¾è±¡ è©•ä¾¡æŒ‡æ¨™ Inception Score Human Rank å¤šæ§˜æ€§ã¨ç”»è³ªã«é–¢ã™ã‚‹è©•ä¾¡ è¨è¨ˆã®å¦¥å½“æ€§è©•ä¾¡ Stageã®æ®µéšŽåˆ†ã‘ ã‚¬ã‚¦ã‚¹åˆ†å¸ƒã«ã‚ˆã‚‹å…¥åŠ›ç”Ÿæˆ å…¥åŠ›Textã¸ã®å¯¾å¿œè©•ä¾¡ çµè«– å®Ÿè£… æ„Ÿæƒ³ è«–æ–‡ [1612.03242] StackGAN: Text to Photo-realistic Image Synthesis with Stacked Generative Adversarial Networks è‘—è€… Han Zhang, Tao Xu, Hongsheng Li, Shaoting Zhang, Xiaogang Wang,
proto_jp 2022/08/12
ai

GAN

image

Midjourney
ãƒªãƒ³ã‚¯
Generative Adversarial Networks (GAN) ã¨ã¯ä½•ã‹ã€ã©ã‚“ãªãƒ–ãƒ¬ãƒ¼ã‚¯ã‚¹ãƒ«ãƒ¼ãŒæœŸå¾…ã§ãå¦‚ä½•ãªã‚‹å±é™ºæ€§ãŒã‚ã‚‹ã‹
Generative Adversarial Networks (GAN) ã¨ã¯Google Brain (AIç ”ç©¶éƒ¨é–€) ã®Ian GoodfellowãŒé–‹ç™ºã—ãŸãƒ‹ãƒ¥ãƒ¼ãƒ©ãƒ«ãƒãƒƒãƒˆãƒ¯ãƒ¼ã‚¯ã§ã€ãã®æ½œåœ¨èƒ½åŠ›ã«æœŸå¾…ãŒé«˜ã¾ã‚Šç ”ç©¶é–‹ç™ºãŒé€²ã‚“ã§ã„ã‚‹ã€‚GANã¯æ§˜ã€…ãªãƒãƒªã‚¨ãƒ¼ã‚·ãƒ§ãƒ³ãŒã‚ã‚Šå¤šå½©ãªæ©Ÿèƒ½ã‚’æŒã£ã¦ã„ã‚‹ã€‚GANã¯Deep LearningãŒæŠ±ãˆã‚‹å•é¡Œã‚’è§£æ±ºã™ã‚‹ç³¸å£ã«ãªã‚‹ã¨è¦‹ã‚‰ã‚Œã¦ã„ã‚‹ã€‚åŒæ™‚ã«ã€GANã¯å¤§ããªå±é™ºæ€§ã‚’å†…åœ¨ã—æ³¨æ„ã‚’è¦ã™ã‚‹æŠ€æ³•ã§ã‚‚ã‚ã‚‹ã€‚ DCGANï¼šç¾å®Ÿãã£ãã‚Šã®ã‚¤ãƒ¡ãƒ¼ã‚¸ã‚’ãƒªã‚¢ãƒ«ã«ç”Ÿæˆ GANã¯ãƒ‹ãƒ¥ãƒ¼ãƒ©ãƒ«ãƒãƒƒãƒˆãƒ¯ãƒ¼ã‚¯ã®æŠ€æ³•ã§äºŒã¤ã®å¯¾å³™ã™ã‚‹ (Adversarial) ãƒãƒƒãƒˆãƒ¯ãƒ¼ã‚¯ãŒã‚³ãƒ³ãƒ†ãƒ³ãƒ„ (ã‚¤ãƒ¡ãƒ¼ã‚¸ã‚„éŸ³å£°ãªã©) ã‚’ç”Ÿæˆã™ã‚‹ã€‚GANã¯æŠ€æ³•ã®ç·ç§°ã§å¤šãã®ãƒãƒªã‚¨ãƒ¼ã‚·ãƒ§ãƒ³ãŒã‚ã‚‹ã€‚ãã®ä»£è¡¨ãŒDCGAN (Deep Convolutional Generative Adversarial Netw
proto_jp 2022/08/12
ai

image

GAN

Midjourney
ãƒªãƒ³ã‚¯
Self-attention GANï¼ˆSAGANï¼‰ã‚’å®Ÿè£…ã—ã¦éŠã‚“ã§ã¿ãŸ
å‰å›žã®æŠ•ç¨¿ã§ã¯ã€Spectral Noramlizationã‚’ä½¿ã£ãŸGANã€ŒSNGANã€ã‚’å®Ÿè£…ã—ã¾ã—ãŸãŒã€ãã‚Œã®å¿œç”¨ç³»ã§ã‚ã‚‹Self-attention GANã€ŒSAGANã€ã‚’å®Ÿè£…ã—ã¦éŠã‚“ã§ã¿ã¾ã—ãŸã€‚CIFAR-10ã€STL-10ã€AnimeFace Datasetã€Oxford Flowerã‚’ç”Ÿæˆã—ã¦ç¢ºã‹ã‚ã¦ã„ã¾ã™ã€‚ ï¼ˆé•·ã„ã‚“ã§æ°—æ¥½ã«èªã‚“ã§ãï¼‰ Self-attention GANã¨ã¯ è«–æ–‡ï¼šSelf-Attention Generative Adversarial Networks https://arxiv.org/abs/1805.08318 ä¸»ãªç‰¹å¾´ åŸºæœ¬çš„ãªç™ºæƒ³ã¯Spectral Noramlizationã‚’ä½¿ã£ãŸGANï¼ˆSNGANï¼‰ã®ç™ºå±•ã€‚SNGANã®è«–æ–‡ã€è‡ªåˆ†ã®è¨˜äº‹ GANã«å®‰å®šæ€§ã«å¤§ããè²¢çŒ®ã—ãŸSpectral Noramlizationã‚’ã€SNGANã¯Dï¼ˆDiscrim
proto_jp 2022/08/12
ai

deeplearning

image

GAN
ãƒªãƒ³ã‚¯
Generative Adversarial Networkï¼ˆGANï¼‰ã‚’ä½¿ã£ãŸç”»åƒç”Ÿæˆã®é€²åŒ–ã‚’ç´¹ä»‹
ç¾åœ¨ä¸–é–“ã‚’è³‘ã‚ã—ã¦ã„ã‚‹AI æŠ€è¡“ã¨è¨€ãˆã°Deep Learningã§ã™ã‚ˆãã€‚Deep Learningã§ç”»åƒã‚’ç”Ÿæˆã§ãã‚‹ã¨ã„ã†äº‹ã¯åº¦ã€…ãƒ‹ãƒ¥ãƒ¼ã‚¹ç‰ã§å–ã‚Šä¸Šã’ã‚‰ã‚Œã‚‹ã“ã¨ã‚‚ã‚ã‚Šã€å®Ÿéš›ã«ç›®ã«ã•ã‚ŒãŸäº‹ãŒã‚ã‚‹æ–¹ã‚‚ã„ã‚‰ã£ã—ã‚ƒã‚‹ã“ã¨ã§ã—ã‚‡ã†ã€‚ ã“ã†ã—ãŸãƒ‹ãƒ¥ãƒ¼ã‚¹ãŒå‡ºã¦ãã‚‹ã‚ˆã†ã«ãªã£ãŸèƒŒæ™¯ã«ã¯ã€2014å¹´ã«ç™ºè¡¨ã•ã‚ŒãŸGenerative Adversarial Networkï¼ˆGANï¼‰ã¨ã„ã†ã‚¢ã‚¤ãƒ‡ã‚¢ãŒæ·±ãé–¢ä¿‚ã—ã¦ãŠã‚Šã€æ§˜ã€…ãªç ”ç©¶ã®æˆæžœã‚‚ã‚ã£ã¦ã“ã®æŠ€è¡“ã¯æœ€åˆã«ç™ºè¡¨ã•ã‚Œã¦ã‹ã‚‰ãŸã£ãŸï¼•å¹´ã§å‡„ã¾ã˜ã„é€²åŒ–ã‚’é‚ã’ã¦ã„ã¾ã™ã€‚ ä¾‹ãˆã°ã€ This Person Does Not Exist ã¨ã„ã†ã‚µã‚¤ãƒˆã«ã‚¢ã‚¯ã‚»ã‚¹ã™ã‚‹ã¨äººç‰©ã®ç”»åƒãŒè¡¨ç¤ºã•ã‚Œã¾ã™ãŒã€ã“ã‚ŒãŒç¾å®Ÿã«ã¯å˜åœ¨ã—ãªã„äººã ã¨èžã„ã¦é©šã‹ã‚Œã‚‹æ–¹ã¯å¤šã„ã‹ã‚‚ã—ã‚Œã¾ã›ã‚“ã€‚ ã“ã®ã‚µã‚¤ãƒˆã«è¡¨ç¤ºã•ã‚Œã‚‹äººç‰©ç”»åƒã¯å…¨ã¦Deep Learningã«ã‚ˆã‚‹ã‚‚ã®ã§ã€äººç‰©ç”»åƒã‚’ãŠæ‰‹æœ¬ã«å¦ç¿’ã•ã›ã€ãŠæ‰‹æœ¬ã¨è¦‹
proto_jp 2022/08/11
GAN

deeplearning

image
ãƒªãƒ³ã‚¯
1