【自己幫影片換主角】FakeApp 教學

留名

有成品未

知唔知檔案總管有搜尋

終於搞店 搵太多相成日都out of memory
一放入去已經係0.053同0.058

有人整fatboyeatallday我派500蚊利是比佢

巴打
係咪我部機太廢
GTX 750ti
i7-4790
16 RAM

batch size要大過14

個個唔同 至少我自己係放番落先drive先run到

有OOM多數都係train個到啲數太大

我Extract出黎既係jpeg/jpg
位置有無錯？最尾係咪aligned?

睇黎鬼妹轉asian都唔係好得

一轉日本妹改香港人 就即刻0.01

屌我個明步啊 本身run左成日已經0.03 唔知點解一返左黎由0開始

lm

終於整到第一條片
但成效好差
睇黎要瘋狂訓練多啲先得

location肯定無錯
重裝過CUDA8.0都唔得

Using TensorFlow backend.
Traceback (most recent call last):
File "train.py", line 97, in <module>
File "utils.py", line 14, in load_images
MemoryError
[4540] Failed to execute script train

C:\Users\com\Desktop\FakeApp>

想問係咪ram問題

Png

果解我extract前素材係png但出完aligned既都係jpeg

但我fakeapp file type係png

我Extract出黎既係jpeg/jpg
位置有無錯？最尾係咪aligned?

location肯定無錯
重裝過CUDA8.0都唔得

之前我都遇過呢個話圖唔得嘅問題
之後我將

Model Path: C:/fakes/model
Data A Path: C:/fakes/data_A/aligned
Data B Path: C:/fakes/data_B/aligned

改做

Model Path: c:/fakes/model
Data A Path: c:/fakes/data_A/aligned
Data B Path: c:/fakes/data_B/aligned

就解決左

Cccc

去到呢到收左皮 Train: 即係比套軟件學習
8432 failed to execute script train

去到呢到收左皮 Train: 即係比套軟件學習
8432 failed to execute script train

copy晒成段野出黎先知咩事