Multi-Scale Efficient Global Context Vision Transformer (MS-EffGCViT) is a hybrid CNN-ViT architecture for deepfake detection. It fuses CNN-driven spatial inductive bias with hierarchical global-context attention to catch both local artifacts (textures, blending seams) and global artifacts (lighting, structural inconsistency). A single architecture ships in two sizes and three domain-tuned checkpoints, working on both static images and video at the frame level. - Frame-level — one model handles both images and videos (frame-level inference + aggregation). - Cross-domain — robust on both East-Asian (KoDF) and Western (Celeb-DF-v2, FaceForensics++) faces. - Two variants — Fast (b0) for…
Open-weight model · Video classification
ms-eff-gcvit-deepfake-b5-ff-plus-plus
by YUNJE SEO KoreaPeter/ms-eff-gcvit-deepfake-b5-ff-plus-plus
ms-eff-gcvit-deepfake-b5-ff-plus-plus is an open-weight model for video classification from YUNJE SEO, released under MIT License. It has 53M parameters. At 16-bit it needs about 0.1 GB of GPU memory, which fits on 1x MI300X from $1.85 an hour, at the lowest prices in the SAVRN Index. It draws 6.6k downloads a month.
Multi-Scale Efficient Global Context Vision Transformer (MS-EffGCViT) is a hybrid CNN-ViT architecture for deepfake detection.
Parameters53M
Context—
Weights226.5 MB
Licensemit
AccessOpen weights
Monthly Downloads6.6k
Runs On
What it takes to serve ms-eff-gcvit-deepfake-b5-ff-plus-plus (53M parameters): the memory its weights need at each precision, and the cheapest way to rent enough data-center GPUs to hold them.
| Precision | Weights | Memory needed | Cheapest setup | Per hour | Also fits |
|---|---|---|---|---|---|
| 16-bit | 0.1 GB | 0.1 GB | 1x MI300X (192 GB) Vultr |
$1.85 | 1x H100 $1.99 · 1x MI325X $2.00 |
| 8-bit | 0.1 GB | 0.1 GB | 1x MI300X (192 GB) Vultr |
$1.85 | 1x H100 $1.99 · 1x MI325X $2.00 |
| 4-bit | 0.0 GB | 0.0 GB | 1x MI300X (192 GB) Vultr |
$1.85 | 1x H100 $1.99 · 1x MI325X $2.00 |
Memory is the weights at that precision plus 20% for the runtime and a short context; a long context needs more. Prices are the lowest on-demand hourly rates in the SAVRN Index, read Oct 9, 2026.
Model Card
By YUNJE SEO, published under mit, revision a8fd324d86ef.
Multi Scale Efficient Global Context Vision Transformer
GitHub Repository: HanMoonSub/DeepGuard
Live demo: DeepFake Video Detection
Live demo: DeepFake Image Detection
Live demo: DeepFake Detection XAI
Multi-Scale Efficient Global Context Vision Transformer (MS-EffGCViT) is a hybrid CNN-ViT architecture for deepfake detection. It fuses CNN-driven spatial inductive bias with hierarchical global-context attention to catch both local artifacts (textures, blending seams) and global artifacts (lighting, structural inconsistency).
A single architecture ships in two sizes and three domain-tuned checkpoints, working on both static images and video at the frame level.
Core Features
- Frame-level — one model handles both images and videos (frame-level inference + aggregation).
- Cross-domain — robust on both East-Asian (KoDF) and Western (Celeb-DF-v2, FaceForensics++) faces.
- Two variants — Fast (b0) for real-time/edge, Pro (b5) for enterprise accuracy.
- timm-compatible — load via the
timminterface or thedeepguardpackage.
Model Specifications
Configuration
- Architecture
- MsEffGCViTForImageClassification
- Model type
- ms_eff_gcvit
Identity and Version
- Repository
- KoreaPeter/ms-eff-gcvit-deepfake-b5-ff-plus-plus
- Publisher
- YUNJE SEO
- Task
- Video classification
- Modality
- Video
- Library
- transformers
- Parameters
- 53M parameters
- Languages
- Not stated by the source
- Revision
- a8fd324d86ef7e4caeb2a693e1207d794be32b9f
- First published
- 2026-06-23
- Last updated
- 2026-07-04
Files and Weights
12 files, 229.0 MB in total. The weights are 2 files totalling 226.5 MB in pt, safetensors.
Weights2 files · 226.5 MB
Configuration5 files · 13.2 KB
Documentation1 file · 11.4 KB
Other3 files · 2.5 MB
Repository1 file · 1.6 KB
Every file
| File | Type | Size | SHA-256 |
|---|---|---|---|
| model.safetensors | Weights | 220.1 MB | 843a63c3359d |
| yolov8n-face.pt | Weights | 6.4 MB | d545bf1add5a |
| config.json | Configuration | 1.5 KB | — |
| configuration_ms_eff_gcvit.py | Configuration | 1.5 KB | — |
| modeling_ms_eff_gcvit.py | Configuration | 1.3 KB | — |
| pipeline_ms_eff_gcvit.py | Configuration | 3.5 KB | — |
| pipeline_video_ms_eff_gcvit.py | Configuration | 5.4 KB | — |
| README.md | Documentation | 11.4 KB | — |
| dual_branch.gif | Other | 2.3 MB | 3b17745fc2ef |
| ms_eff_gcvit.JPG | Other | 100.3 KB | 9cc6e353e563 |
| window_attention.JPG | Other | 45.2 KB | — |
| .gitattributes | Repository | 1.6 KB | — |
License and Download
- License
- mit
- Access
- Open weights, no gate
- Download size
- 226.5 MB
Download from YUNJE SEO
Released by YUNJE SEO through its official repository on Hugging Face. Read the license.
Built From
- Derived from timm/tf_efficientnet_b5.ns_jft_in1k
- Described by arXiv:1901.08971
- Trained on (disclosed) ILSVRC/imagenet-1k
Memory Requirements
| Precision | Weights in memory |
|---|---|
| As published | 226.5 MB |
| 16-bit | 0.1 GB |
| 8-bit | 0.1 GB |
| 4-bit | 0.0 GB |
Weights only, from the published parameter count; the key-value cache and runtime add to this.
Questions About ms-eff-gcvit-deepfake-b5-ff-plus-plus
How much GPU memory does ms-eff-gcvit-deepfake-b5-ff-plus-plus need?
About 0.1 GB at 16-bit and 0 GB at 4-bit: the weights (53M parameters) plus a working margin. A long context needs more.
What is the cheapest GPU to run ms-eff-gcvit-deepfake-b5-ff-plus-plus on?
At 16-bit, 1x MI300X from $1.85 an hour; at 4-bit, 1x MI300X from $1.85 an hour, at the lowest on-demand prices the SAVRN Index lists.
Can I use ms-eff-gcvit-deepfake-b5-ff-plus-plus commercially?
Yes. ms-eff-gcvit-deepfake-b5-ff-plus-plus is released under MIT License. The MIT License is a short permissive license. It permits commercial use, modification and redistribution, provided the copyright notice and permission notice are included.
Similar Models
Multi-Scale Efficient Global Context Vision Transformer (MS-EffGCViT) is a hybrid CNN-ViT architecture for deepfake detection. It fuses CNN-driven spatial inductive bias with hierarchical global-context attention to catch both local artifacts (textures, blending seams) and global artifacts (lighting, structural inconsistency). A single architecture ships in two sizes and three domain-tuned checkpoints, working on both static images and video at the frame level. - Frame-level — one model handles both images and videos (frame-level inference + aggregation). - Cross-domain — robust on both East-Asian (KoDF) and Western (Celeb-DF-v2, FaceForensics++) faces. - Two variants — Fast (b0) for…
VideoMAEv2-Base model pre-trained for 800 epochs in a self-supervised way on UnlabeldHybrid-1M dataset. It was introduced in the paper [[CVPR23]VideoMAE V2: Scaling Video Masked Autoencoders with Dual Masking](https://arxiv.org/abs/2203.12602) by Wang et al. and first released in GitHub. You can use the raw model for video feature extraction. Here is how to use this model to extract a video feature
Model · Video classification
videomae-base-finetuned-kinetics-violence-nonviolence-tuned
This model is a fine-tuned version of MCG-NJU/videomae-base-finetuned-kinetics on an unknown dataset. It achieves the following results on the evaluation set: The following hyperparameters were used during training: - learningrate: 5e-05 - trainbatchsize: 16 - evalbatchsize: 16 - lrschedulertype: linear - lrschedulerwarmupratio: 0.1 - trainingsteps: 348 - Transformers 4.49.0 - Pytorch 2.6.0+cu126 - Datasets 3.3.2 - Tokenizers 0.21.0
This model is a fine-tuned version of MCG-NJU/videomae-base for binary violence classification (violent / non-violent). It builds on Nikeytas/videomae-crime-detector-production-v1, which was itself fine-tuned from videomae-base on a subset of UCF Crime. Starting from that checkpoint, this model was further fine-tuned on the Bus Violence Dataset to close the domain gap to public-transport surveillance footage. - UCF Crime (jinmang2/ucfcrime) — inherited from the base checkpoint - Bus Violence Dataset (Zenodo) — real moving-bus footage, binary violent / non-violent labels, used for domain-specific fine-tuning Evaluated on a held-out Bus Violence Dataset test split (n = 280). The base…
This model is a fine-tuned version of MCG-NJU/videomae-base on an unknown dataset. It achieves the following results on the evaluation set: The following hyperparameters were used during training: - learningrate: 5e-05 - trainbatchsize: 8 - evalbatchsize: 8 - lrschedulertype: linear - trainingsteps: 370 - Transformers 5.16.1 - Pytorch 2.14.0+cu126 - Datasets 5.0.1 - Tokenizers 0.23.2