Update README.md

README.md

@@ -17,29 +17,21 @@ Custom trained models for the soon to be released **Forbidden Vision** ComfyUI c
 
 ---
 
-## 📦 Models
+Custom-trained models for high-performance face detection, landmark estimation, and segmentation across realistic, anime, and NSFW content.
 
-All models were trained on a custom dataset of **10k+ images** (a mix of SFW/NSFW from Civitai, Danbooru, real photos and my own curated images).  
-The goal was to cover style diversity across anime and realistic domains.
+🎯 What Makes These Different?
+Trained on what you actually generate.
+Unlike general-purpose face models trained only on real photographs, these models were specifically trained on:
 
-### 🟢 Face Detection (YOLO 11 small)
-- **Base:** YOLO face detection models   
-- **Variants:**  
-  - `ForbiddenVision_facedetect_yolov11s.pt` → YOLO Small fine-tune -> generally more robust and reliable
-  - `ForbiddenVision_facedetect_yolov11n.pt` → YOLO Nano fine-tune -> if you need maximum speed and lower vram usage
-- **Training:** Fine-tuned on the <a href="https://github.com/YapaLab/yolo-face">YOLOv11 Face base</a>.  
-- **Usage:** Lightweight, accurate bounding box detection for faces.  
+Diffusion outputs from CivitAI (realistic and anime styles)
+Danbooru anime dataset
+Real photography for ground truth accuracy
+NSFW content without artificial restrictions
 
----
-
-### 🔵 Face Segmentation (EfficientNet-B0 advprop)
-- **Base:** timm EfficientNet-B0   
-- **Training:** Besides manually annotated civitai and custom images, the model used an equal split of <a href="https://github.com/switchablenorms/CelebAMask-HQ">CelebAMask-HQ</a>
-- **Usage:** Provides clean masks for downstream processing.  
-
----
+This mixed training approach ensures reliable performance across:
 
-### 🟣 Landmark Rotation (MobileNetV3-Small)
-- **Base:** MobileNetV3-Small 
-- **Training:** Trained for landmark-based rotation correction.  
-- **Usage:** Aligns faces to level eyes for consistent preprocessing.
+✓ Both realistic and anime art styles
+✓ Difficult angles, occlusions, and expressions
+✓ Low-quality generations and artifacts
+✓ SFW and NSFW content equally
+✓ Edge cases that break traditional face detectors