Hi everyone 👋 I’m new to Zindi competitions and excited to be learning from the community.

I’m currently a bit stuck and would really appreciate some guidance on model choice and approach. My leaderboard score is stuck around 0.83, and I’m struggling to push beyond that.

What I’ve tried so far:

• DINOv2 (frozen backbone): CLS token + simple regression heads Patch tokens with attention-based pooling for feature aggregation
• ConvNeXt with a regression heads
• Target scaling / normalization

Unfortunately, none of these approaches have helped me improve beyond the current LB score. I’d appreciate any advice on what I should consider doing next in my experiments