GAN Compression Efficient Architectures for Interactive Conditional GANs

GAN Compression : Efficient Architectures for Interactive Conditional GANs

Conditional Generative Adversarial Networks (cGANs) have enabled controllable image synthesis for many vision and graphics applications. However, recent cGANs are 1-2 orders of magnitude more compute-intensive than modern recognition CNNs. For example, GauGAN consumes 281G MACs per image, compared t...

Ausführliche Beschreibung

Bibliographische Detailangaben
Veröffentlicht in:IEEE transactions on pattern analysis and machine intelligence. - 1979. - 44(2022), 12 vom: 09. Dez., Seite 9331-9346
1. Verfasser: Li, Muyang (VerfasserIn)
Weitere Verfasser: Lin, Ji, Ding, Yaoyao, Liu, Zhijian, Zhu, Jun-Yan, Han, Song
Format: Online-Aufsatz
Sprache:English
Veröffentlicht: 2022
Zugriff auf das übergeordnete Werk:IEEE transactions on pattern analysis and machine intelligence
Schlagworte:Journal Article
LEADER 01000naa a22002652 4500
001 NLM332903826
003 DE-627
005 20231225220532.0
007 cr uuu---uuuuu
008 231225s2022 xx |||||o 00| ||eng c
024 7 |a 10.1109/TPAMI.2021.3126742  |2 doi 
028 5 2 |a pubmed24n1109.xml 
035 |a (DE-627)NLM332903826 
035 |a (NLM)34752389 
040 |a DE-627  |b ger  |c DE-627  |e rakwb 
041 |a eng 
100 1 |a Li, Muyang  |e verfasserin  |4 aut 
245 1 0 |a GAN Compression  |b Efficient Architectures for Interactive Conditional GANs 
264 1 |c 2022 
336 |a Text  |b txt  |2 rdacontent 
337 |a ƒaComputermedien  |b c  |2 rdamedia 
338 |a ƒa Online-Ressource  |b cr  |2 rdacarrier 
500 |a Date Revised 08.11.2022 
500 |a published: Print-Electronic 
500 |a Citation Status PubMed-not-MEDLINE 
520 |a Conditional Generative Adversarial Networks (cGANs) have enabled controllable image synthesis for many vision and graphics applications. However, recent cGANs are 1-2 orders of magnitude more compute-intensive than modern recognition CNNs. For example, GauGAN consumes 281G MACs per image, compared to 0.44G MACs for MobileNet-v3, making it difficult for interactive deployment. In this work, we propose a general-purpose compression framework for reducing the inference time and model size of the generator in cGANs. Directly applying existing compression methods yields poor performance due to the difficulty of GAN training and the differences in generator architectures. We address these challenges in two ways. First, to stabilize GAN training, we transfer knowledge of multiple intermediate representations of the original model to its compressed model and unify unpaired and paired learning. Second, instead of reusing existing CNN designs, our method finds efficient architectures via neural architecture search. To accelerate the search process, we decouple the model training and search via weight sharing. Experiments demonstrate the effectiveness of our method across different supervision settings, network architectures, and learning methods. Without losing image quality, we reduce the computation of CycleGAN by 21×, Pix2pix by 12×, MUNIT by 29×, and GauGAN by 9×, paving the way for interactive image synthesis 
650 4 |a Journal Article 
700 1 |a Lin, Ji  |e verfasserin  |4 aut 
700 1 |a Ding, Yaoyao  |e verfasserin  |4 aut 
700 1 |a Liu, Zhijian  |e verfasserin  |4 aut 
700 1 |a Zhu, Jun-Yan  |e verfasserin  |4 aut 
700 1 |a Han, Song  |e verfasserin  |4 aut 
773 0 8 |i Enthalten in  |t IEEE transactions on pattern analysis and machine intelligence  |d 1979  |g 44(2022), 12 vom: 09. Dez., Seite 9331-9346  |w (DE-627)NLM098212257  |x 1939-3539  |7 nnns 
773 1 8 |g volume:44  |g year:2022  |g number:12  |g day:09  |g month:12  |g pages:9331-9346 
856 4 0 |u http://dx.doi.org/10.1109/TPAMI.2021.3126742  |3 Volltext 
912 |a GBV_USEFLAG_A 
912 |a SYSFLAG_A 
912 |a GBV_NLM 
912 |a GBV_ILN_350 
951 |a AR 
952 |d 44  |j 2022  |e 12  |b 09  |c 12  |h 9331-9346