search

NasimShafiee / Introvert-Human-trajectory-prediction-via-conditional-3d-attention

Human trajectory prediction via conditional 3d attention
7 stars 3 forks source link

Weight of the model #2

Open while-lin opened 1 year ago

while-lin commented 1 year ago

Hi, thank you very much for your nice work and sharing it!

Could You share the weight of the model on ETH/UCY? Thinks!

while-lin commented 1 year ago

AND I have another issues In the dataset you give,I guess a pedestrian's trajectory may be used for training several times. image IN the pic above, Ped have 22 timesteps in the video, and the 22 timesteps Produced three slice, which is (0~20),(1~21),and (2~22), IS that ok?

NasimShafiee commented 1 year ago

Each datapoint contains 20 timesteps, 8 observations and 12 predictions.

Get Outlook for iOShttps://aka.ms/o0ukef

From: while-lin @.> Sent: Saturday, October 29, 2022 10:45:17 AM To: NasimShafiee/Introvert-Human-trajectory-prediction-via-conditional-3d-attention @.> Cc: Subscribed @.***> Subject: Re: [NasimShafiee/Introvert-Human-trajectory-prediction-via-conditional-3d-attention] Weight of the model (Issue #2)

AND I have another issues In the dataset you give,I guess a pedestrian's trajectory may be used for training several times. [image]https://nam12.safelinks.protection.outlook.com/?url=https%3A%2F%2Fuser-images.githubusercontent.com%2F52849152%2F198845616-8fa77dcd-e99f-4587-8ffd-acd4a9fd984c.png&data=05%7C01%7Cshafiee.n%40northeastern.edu%7Cb8057795593b4951e67d08dab9d557a7%7Ca8eec281aaa34daeac9b9a398b9215e7%7C0%7C0%7C638026623205033371%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=uSqTdgkoq0dvDe0q%2F1bI3itykaRmyGxkcGkYYwpHtz8%3D&reserved=0 IN the pic above, Ped have 22 timesteps in the video, and the 22 timesteps Produced three slice, which is (020),(121),and (2~22), IS that ok?

— Reply to this email directly, view it on GitHubhttps://nam12.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2FNasimShafiee%2FIntrovert-Human-trajectory-prediction-via-conditional-3d-attention%2Fissues%2F2%23issuecomment-1295915591&data=05%7C01%7Cshafiee.n%40northeastern.edu%7Cb8057795593b4951e67d08dab9d557a7%7Ca8eec281aaa34daeac9b9a398b9215e7%7C0%7C0%7C638026623205189607%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=Lh2QGuhczXBFuYPnpYcKSzX2Mnz%2BWFn934k20mPLsd0%3D&reserved=0, or unsubscribehttps://nam12.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fnotifications%2Funsubscribe-auth%2FAHSMRT4VQFC722WAMUMBSTTWFVPC3ANCNFSM6AAAAAAQ6DYVBU&data=05%7C01%7Cshafiee.n%40northeastern.edu%7Cb8057795593b4951e67d08dab9d557a7%7Ca8eec281aaa34daeac9b9a398b9215e7%7C0%7C0%7C638026623205189607%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=qaQrirquvJkU%2Bq8ol5HFBexn%2FFLu9X0fAkwEstb8vrA%3D&reserved=0. You are receiving this because you are subscribed to this thread.Message ID: @.***>

while-lin commented 1 year ago

yes i know that. Well, I mean, the pos_xy of (1-20)timesteps in data_id0 is euqal to the pos_xy of (0-19)timesteps in data_id1, and I don't know whether it is reasonable to split the row data in this way