dreamerlin
commented
2 years ago @17Skye17 It seems that the author will remove '-' and '_' chars when a video id starts with these two.
And @sallymmx Could you please provide the validation videos ?
Open 17Skye17 opened 2 years ago
dreamerlin
commented
2 years ago @17Skye17 It seems that the author will remove '-' and '_' chars when a video id starts with these two.
And @sallymmx Could you please provide the validation videos ?
sallymmx
commented
2 years ago 您好，我是王蒙蒙，您的邮件我已收到，并及时回复。
dreamerlin
commented
2 years ago 您好，我是王蒙蒙，您的邮件我已收到，并及时回复。
谢谢，请问您能提供这些 video_id 的视频吗，可以发到 jintaolin@smail.nju.edu.cn
NtUqv_6vdU
3MhOA-vSO8
JwMoMeZ_v8
LrBC1_yf04
c96bD-9fHs
blq_c14hGQ
bVVs-nntQ
sSTHZHHp-c
WI7e5-wURs
iox_MbwxNg
IAo-mNduUk
2L8B_meOLE
wdnasc-fCg
NMaC-IGv_Q
OoJW-OeFtw
jR-X0LqwpY
c-YbuFrXbI
H-Ww0gGDWU
Xfmjzt_n24
I3_52Xh7oU
UqI-TBQRgg
RlFMUo-JE
Ykfa-4qx0
dguKqz_F5w
2DwBhMUH-I
u8zlG-OS_E
afwq-zVgJk
96AfwOj-qw
4SJ-uWc3PQ
esH_aGzBrw
xxBx1jv-ks
Zpj4-Z2YRk
HWxZHHT_l4
G_lySaTeNM
pUDgyU_KGc
GQ-4QfVpXc
D-Fa71ta14
u5A74I0-M
RZXH93_XNY
d8_H5d2sd4
Q_vnBY8YP8
7TZOYU_Ta0
fZ0IO-Q-ZQ
FawHl2-DAg
t_T_nYKdh8
7Od7A1-B9s
rQuS0w-1b8
n0WAbM8z-o
spJJybwq-g
OnU1Hr_jlY
LKhtbW5q-c
kUfOBKJZczk
vYfm8bO-TM
J-8cbYBG7c
a06xpsZj-U
uVv1h-xAe0
B_pr_4s7vY
YGwB8HJj-g
VP-VaZhno0
r5c12Eo_jY
yAlJ1P_SGg
ITTI-fkvo
GBKRR-OvqM
5Vu8HJ_eMg
D5-ZGEjiWI
06oD_bFxOQ
clhd73_vDQ
IyR-sGt0uw
hR-iDJcjgU
blRZN-6_ZM
G2XYLk0-38
FGj7-Cxu_0
LktSuL8_7M
S6wwANH-EY
gxLOV_s9wU
WRh2_MJLLs
UxN_uuEZC0
TyWiE-4zpM
rrtLyJs-3w
TL-9g8KBFA
5vTJ-N4jrI
HUiafDe-xTo
Etym1-30wM
4J-bkpjVb4
7-dud_cqq8
PBf_Wa6vO8
5EvC-g-KUQ
17Skye17
commented
2 years ago @17Skye17 It seems that the author will remove '-' and '_' chars when a video id starts with these two.
And @sallymmx Could you please provide the validation videos ?
yes, it is, after training the model with my own dataset copy I still can't get the reported accuracy, so I was wondering if the validation set is a devil.
sallymmx
commented
1 year ago The validation videos are set like the official K400.
Hi, thank you for releasing the code. After checking the validation file, I found ~180 video ids that can not be found in official val file or some of them can not be decoded, the videos are listed below:
I am not sure if the difference would cause ~1% performance drop in my testing case, since I loaded pretrained model and checked all testing parameters provided in the config file.