Downloads · 30 days
0
findit-dac-2026
darrnhard/findit-dac-2026
findit-dac-2026 is a machine learning model from darrnhard. Use it for the machine learning task on the model card, and read the license before you ship it in a product.
modelcard = """--- license: mit tags: - face-anti-spoofing - image-classification - timm - pytorch ---
Access
Public
Updated Apr 26, 2026
Repo size
9 GB
Likes
0
Public
Repo makeup
Click a slice to open those files.
.pth9 GB · 100%
At a glance
- Access
- Public
- Created
- Apr 22, 2026
- Updated
- Apr 26, 2026
- SHA
- c62a23e1
Model card
From the Hugging Face model README
model_card = """--- license: mit tags:
- face-anti-spoofing
- image-classification
- timm
- pytorch
FindIT-DAC 2026 — Face Anti-Spoofing (exp07)
Checkpoints from the FindIT-DAC 2026 Kaggle competition.
Task: 6-class face anti-spoofing | Metric: Macro F1
Best public LB: 0.79207 (all-5 equal-weight ensemble, argmax)
Architectures (exp07, 5 folds each)
| Model | timm name | OOF F1 |
|---|---|---|
| EVA-02-Base | eva02_base_patch14_224.mim_in22k | 0.9293 |
| DINOv2-Base | vit_base_patch14_reg4_dinov2.lvd142m | 0.9254 |
| SwinV2-Base | swinv2_base_window12_192.ms_in22k | 0.9131 |
| ConvNeXt-Base | convnext_base.fb_in22k | 0.9073 |
| EfficientNet-B4 | efficientnet_b4 | 0.8826 |
File layout
models/exp07/<arch>/fold_<0-4>/best_model.pth
oof/exp07/ # OOF prediction arrays per architecture
Source code
github.com/darrnhard/findit-dac-2026 """
with open("hf_model_card.md", "w") as f: f.write(model_card)
api.upload_file( path_or_fileobj="hf_model_card.md", path_in_repo="README.md", repo_id=REPO_ID, repo_type="model", )
---
- Created
- Apr 22, 2026
- Updated
- Apr 26, 2026