Ir para o conteúdo
ModelListadosExterno
Browse files

CSIRO AUX MODEL

This model card covers the main and auxiliary models used in the Kaggle CSIRO Pasture Biomass Prediction competition (4th place). The solution uses a ViT-Huge DINOv3 backbone with multi-modal fusion and an auxiliary...

Fontehuggingface.co/jatinmehra/CSIRO-AUX_MODEL/tree/608dbe4abe866bfc5ef6cd5f48fd959d1cf6d6ce
TipoOtherModelo basefacebook/dinov3-vith16plus-pretrain-lvd1689mVersãoBest aux only seed44 fold0Cnota

Fonte do modelo

Tipo
Other
Tamanho do arquivo
3,13 GB
Modelo base
facebook/dinov3-vith16plus-pretrain-lvd1689m
Versão
vBest aux only seed44 fold0
Atualizado
2 de nov. de 2026
Fonte
Hugging Face

Descrição da fonte

CSIRO Pasture Biomass Prediction (4th Place Solution)

GitHub Kaggle

This model card covers the main and auxiliary models used in the Kaggle CSIRO Pasture Biomass Prediction competition (4th place). The solution uses a ViT-Huge DINOv3 backbone with multi-modal fusion and an auxiliary pretraining stage.

Model Description

Main model (Stage 1)

  • Backbone: vit_huge_plus_patch16_dinov3.lvd1689m (DINOv3)
  • Input: RGB pasture image + tabular features (Pre_GSHH_NDVI, Height_Ave_cm)
  • Output: Biomass regression targets
  • Loss: Weighted Smooth L1 with weights [0.1, 0.1, 0.1, 0.2, 0.5]

Aux model (Stage 2)

  • Input: RGB pasture image only
  • Output: Predicts tabular features (NDVI, Height)
  • Purpose: Feature enrichment for the main model

Model Weights

Intended Use

  • Research and benchmarking for pasture biomass regression from aerial/ground images.
  • Demonstration of multi-modal fusion and auxiliary prediction benefits.

Out-of-Scope Use

  • Medical or safety-critical applications.
  • Real-time decision systems without domain validation.

Training Data

  • Kaggle CSIRO Pasture Biomass Prediction dataset.
  • Images were manually cropped to remove cardboard background.

Preprocessing

  • Manual cropping to remove cardboard borders.
  • Tabular features and regression targets normalized via StandardScaler.
  • Image size: 800×800.

Training Configuration

Stage 1 (Main Model)

  • Batch size: 10
  • Optimizer: AdamW
  • LR: 5e-5
  • Scheduler: CosineAnnealingWarmRestarts (T_0=10, T_mult=2, eta_min=1e-6)
  • 5-fold CV (Seed 42)

Stage 2 (Aux Model)

  • Batch size: 8
  • Optimizer: AdamW
  • LR: 5e-5
  • Scheduler: ReduceLROnPlateau (factor=0.5, patience=4)
  • 5-fold CV (Seed 44)

Results

StagePublic LBPrivate LB
Baseline0.740.64
+ Data Cleaning0.750.65
+ Auxiliary Training0.760.66

Limitations

  • Trained on competition data with specific capture setup; generalization to other pasture imagery may vary.
  • Large ViT-Huge backbone is compute-intensive.

Ethical Considerations

  • Dataset-specific biases may affect generalization.
  • Manual preprocessing may not be reproducible without the same data access.

Citation

If you use these weights, please cite the Kaggle competition solution.

License

MIT

Fontes

1 fonte
Hugging FacePrimárioVersão 608dbe4abe866bfc5ef6cd5f48fd959d1cf6d6ce
huggingface.co/jatinmehra/CSIRO-AUX_MODEL/tree/608dbe4abe866bfc5ef6cd5f48fd959d1cf6d6ce

Verificado 2 de nov.

Download files

Artefatos de modelo

5 artefatos

Models_Aux_Only_v7/best_aux_only_seed44_fold0.pth

pth · 3,13 GB · SHA-256 5fed35c38587…4e86 · Hugging Face

Download file

Models_Aux_Only_v7/best_aux_only_seed44_fold1.pth

pth · 3,13 GB · SHA-256 19a0c0410747…9041 · Hugging Face

Download file

Models_Aux_Only_v7/best_aux_only_seed44_fold2.pth

pth · 3,13 GB · SHA-256 5ce8decb7d5d…601a · Hugging Face

Download file

Models_Aux_Only_v7/best_aux_only_seed44_fold3.pth

pth · 3,13 GB · SHA-256 7507639cd102…738e · Hugging Face

Download file

Models_Aux_Only_v7/best_aux_only_seed44_fold4.pth

pth · 3,13 GB · SHA-256 1bd7fbf592ee…2ee8 · Hugging Face

Download file

Trechos de fonte

2 trechos
Readme19 de ago.huggingface.co/jatinmehra/CSIRO-AUX_MODEL

--- license: mit datasets: - jatinmehra/CSIRO-Cleaned-Data metrics: - Weighted R2 Scores base_model: - facebook/dinov3-vith16plus-pretrain-lvd1689m pipeline_tag: image-feature-extraction library_name: timm language: - en tags: - agriculture - ComputerVision --- # CSIRO Pasture Biomass Prediction (4th Place Solution) [![GitHub](https://img.shields.io/badge/GitHub-Repository-181717?logo=github&logoColor=white)](https://github.com/Jatin-Mehra119/Kaggle-CSIRO-4th-Position-Solution-) [![Kaggle](https://img.shields.io/badge/Kaggle-4th%20Place%20🥇-gold)](https://www.kaggle.com/competitions/csiro-biomass/writeups/vit-huge-dinov3-and-multi-modal-feature-fusion) This model card covers the **main** and **auxiliary** models used in the Kaggle CSIRO Pasture Biomass Prediction competition (4th place). The solution uses a ViT-Huge DINOv3 backbone with multi-modal fusion and an auxiliary pretraining stage. ## Model Description **Main model (Stage 1)** - Backbone: `vit_huge_plus_patch16_dinov3.lvd1689m` (DINOv3) - Input: RGB pasture image + tabular features (`Pre_GSHH_NDVI`, `Height_Ave_cm`) - Output: Biomass regression targets - Loss: Weighted Smooth L1 with weights `[0.1, 0.1, 0.1, 0.2, 0.5]` **Aux model (Stage 2)** - Input: RGB pasture image only - Output: Predicts tabular features (NDVI, Height) - Purpose: Feature enrichment for the main model ## Model Weights - Main model: https://huggingface.co/jatinmehra/CSIRO-DinoV3-HugePlus-LB76 - Aux model: https://huggingface.co/jatinmehra/CSIRO-AUX_MODEL ## Intended Use - Research and benchmarking for pasture biomass regression from aerial/ground images. - Demonstration of multi-modal fusion and auxiliary prediction benefits. ## Out-of-Scope Use - Medical or safety-critical applications. - Real-time decision systems without domain validation. ## Training Data - Kaggle CSIRO Pasture Biomass Prediction dataset. - Images were manually cropped to remove cardboard background. ## Preprocessing - Manual cropping to remove cardboard borders. - Tabular features and regression targets normalized via `StandardScaler`. - Image size: 800×800. ## Training Configuration **Stage 1 (Main Model)** - Batch size: 10 - Optimizer: AdamW - LR: 5e-5 - Scheduler: CosineAnnealingWarmRestarts (T_0=10, T_mult=2, eta_min=1e-6) - 5-fold CV (Seed 42) **Stage 2 (Aux Model)** - Batch size: 8 - Optimizer: AdamW - LR: 5e-5 - Scheduler: Reduc...

Note19 de ago.huggingface.co/jatinmehra/CSIRO-AUX_MODEL

Source context: 0 downloads · 1 likes · Pipeline image-feature-extraction · Library timm · Repo jatinmehra/CSIRO-AUX_MODEL