Downloads · 30 days
3
13% of all-time downloads
dual-expert-defactify-vit
rijal028/dual-expert-defactify-vit
dual-expert-defactify-vit is a image classification model from rijal028. Use it when you need a label for an image. The card lists the license as apache-2.0.
Arsitektur ensemble spesialis untuk memecahkan classifier bias pada deteksi gambar AI modern. - Pakar Real (Guardian): Mencegah false positive pada foto asli (Weight [3.0, 1.0]). - Pakar AI (Hunter): Menangkap artefak…
All-time downloads
24
Public
Repo size
686 MB
Likes
0
Public
Repo makeup
Click a slice to open those files.
.safetensors686 MB · 100%
At a glance
- Task
- Image Classification
- License
- apache-2.0
- Model type
- dual_expert_vision_transformer
- Access
- Public
- Created
- Sep 19, 2026
- Updated
- Sep 19, 2026
- SHA
- d5ed9080
Model card
From the Hugging Face model README
Dual-Expert Vision Transformer (Defactify Balanced)
Arsitektur ensemble spesialis untuk memecahkan classifier bias pada deteksi gambar AI modern.
- Pakar Real (Guardian): Mencegah false positive pada foto asli (Weight [3.0, 1.0]).
- Pakar AI (Hunter): Menangkap artefak generator difusi modern (Weight [1.0, 3.0]).
Benchmark Evaluasi (500 Sampel Defactify Test)
- Balanced Accuracy: 79.00%
- Macro F1: 79.00%
- Recall Real: 80.00%
- Recall AI: 78.00%
- Latency: ~6.57 ms (Concurrent CUDA Streams)
- Task
- Image Classification
- Type
- dual_expert_vision_transformer
- License
- apache-2.0
- Languages
- en, id
- Created
- Sep 19, 2026
- Updated
- Sep 19, 2026