跳到主要內容

StarGAN

Ref:
https://github.com/yunjey/stargan
https://arxiv.org/pdf/1711.09020.pdf

 StarGAN如圖(b)與其他GAN模型相較, 受到關注的特色在於其不同Domain間的轉換可以使用同一個模型, 不需要一對一Domain的產生很多組模型, 如圖(a).
Picture originated from: https://arxiv.org/pdf/1711.09020.pdf






下圖是使用StarGAN生成的圖片, 一組有六張, 分別是
Input + Black_Hair +Golden_Hair +Brown_Hair +Gender_Change +Aged 共5個Domain

Train這個模型在FX705GE(CPU:INTEL i7-8750H, 32G RAM)上花了約39小時作Training, 總共200,000 steps, 執行速度7 sec/ 10steps ,
相對GL753VE(CPU:INTEL i7-7700HQ, 24G RAM) 90 sec/ 10steps <== 只是worker node, 未開啟NVIDIA CUDA

從下列image看來效果有好有壞, 視人頭比例, training dataset, 相片品質及背景等條件而異.
感謝同仁們(含EX-)及致中提供玉照協助!! 使用celebrity dataset作training 及部份testing



        Input               + Black_Hair       + Golden_Hair    + Brown_Hair       + Gender_Change + Aged






               Input            + Black_Hair       + Golden_Hair     + Brown_Hair     + Gender_Change + Aged





留言

張貼留言

這個網誌中的熱門文章

DeepRacer

Preliminary training: deepracer-github-simapp.tar.gz Reward function: ./opt/install/sagemaker_rl_agent/lib/python3.5/site-packages/markov/environments/deepracer_env.py action = [steering_angle, throttle] TRAINING_IMAGE_SIZE = (160, 120) Plotted waypoints in vertices array of hard track Parameters: on_track, x, y, distance_from_center, car_orientation, progress, steps,                                                                          throttle, steering, track_width, waypoints, closest_waypoints Note: Above picture is from https://yanpanlau.github.io/2016/10/11/Torcs-Keras.html
張貼留言
閱讀完整內容

增強式學習

   迴力球遊戲－ATARI     賽車遊戲DQN－ATARI 賽車遊戲－TORCS Ref:     李宏毅老師 YOUTUBE DRL 1-3 On-policy VS Off-policy On-policy     The agent learned and the agent interacting with the environment is the same     阿光自已下棋學習 Off-policy     The agent learned and the agent interacting with the environment is different     佐助下棋，阿光在旁邊看 Add a baseline:     It is possible that R is always positive     So R subtract a expectation value Policy in " Policy Gradient" means output action, like left/right/fire gamma-discounted rewards: 時間愈遠的貢獻，降低其權重 Reward Function & Action is defined in prior to training MC v.s. TD MC 蒙弟卡羅: critic after episode end : larger variance(cuz conditions differ a lot in every episode), unbiased (judge until episode end, more fair) TD: Temporal-difference approach: critic during episode :smaller variance, biased maybe atari : a3c  ...
張貼留言
閱讀完整內容