Primary competition visual

CGIAR Root Volume Estimation Challenge

Helping Africa
$15 000 USD
Completed (~1 year ago)
Computer Vision
Prediction
1063 joined
257 active
Starti
Jan 24, 25
Closei
Mar 09, 25
Reveali
Mar 10, 25
User avatar
Koleshjr
Multimedia university of kenya
MAJOR SHAKE UP LOADING?
Platform · 3 Mar 2025, 05:54 · 11

I just changed my approach and got a really impressive CV score

0.95xx CV 1.27xx LB

My old approach

1.26xx CV 0.83xx LB

Now I am even more confused 😅. Both are using the same CV structure btw

Discussion 11 answers
User avatar
hashman

Its a seesaw CV :D

3 Mar 2025, 06:26
Upvotes 1
User avatar
CodeJoe

To be frank, my CV has been correlating from the beginning but bad result on the lb. Now I don't know whether I should trust my CV or try another approach🤣😭

3 Mar 2025, 08:33
Upvotes 0
User avatar
MICADEE
LAHASCOM

Smiles... Very true.... CV Score - 0.58 against LB Score - 1.079

3 Mar 2025, 09:47
Upvotes 1
User avatar
CodeJoe

Super 😳

User avatar
CodeJoe

Is your test size that small?

User avatar
Koleshjr
Multimedia university of kenya

Looks like this will be won by luck 😅. 0.58 damnnnnn that's crazy goooood

User avatar
MICADEE
LAHASCOM

In fact, you can say that again, @Koleshjr! That's everyone's prayer right now. This result is really dicey. 😁

User avatar
Koleshjr
Multimedia university of kenya

0.58 with raw targets? I feel like clipping as discussed in the previous discussions can lead to a good score especially if you are clipping both the training and validation targets which is not ideal since the test won't have clipped targets.

User avatar
Koleshjr
Multimedia university of kenya

crazyyy

3 Mar 2025, 13:43
Upvotes 0
User avatar
Ebiendele
Federal university of technology akure

i'm having 0.61 as CV and 1.0244 LB

9 Mar 2025, 17:51
Upvotes 0
User avatar
Koleshjr
Multimedia university of kenya

with raw targets?