feat: 🚀 add YOLOE-11S model deployment for replicate #20

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged

glenn-jocher merged 10 commits into main from feat/yoloe

Aug 7, 2025

.github/workflows/push.yml

-Original file line number
+Diff line change
@@ Expand Up / @@ -15,7 +15,7 @@ jobs: @@
         runs-on: ubuntu-latest
         strategy:
           matrix:
-            model: [yolo11n,yolov8s-worldv2]
+            model: [yolo11n,yolov8s-worldv2,yoloe11s]
         steps:
           - name: Checkout
             uses: actions/checkout@v4
@@ Expand Down @@

.gitignore

-Original file line number
+Diff line change
@@ Expand Up / @@ -182,3 +182,25 @@ weights/ @@
     *_openvino_model/
     *_paddle_model/
     pnnx*
+    *.ts
+    # Replicate deployment folder and files
+    cog.json
+    cog.lock
+    .cog
+    # Ignore Picture files
+    *.jpg
+    *.jpeg
+    *.png
+    *.gif
+    *.bmps
+    # Ignore video files
+    *.mp4
+    *.avi
+    *.mov
+    *.mkv
+    *.webm
+    *.flv
+    *.wmv

yolo11n/predict.py

-Original file line number
+Diff line change
@@ Expand Up / @@ -7,6 +7,8 @@ @@
     class Output(BaseModel):
+        """Output model for predictions."""
         image: Optional[Path] = None
         json_str: Optional[str] = None
@@ Expand Down @@

yoloe11s/README.md

-Original file line number
+Diff line change
@@ -0,0 +1,30 @@
+    # YOLOE-11S Demo Deployment
+    Deploy the official YOLOE-11S model to Replicate with PyTorch inference at <https://replicate.com/ultralytics/yoloe-11s>.
+    ## Setup
+. **Deploy to Replicate:**
+       ```bash
+       cog push r8.im/ultralytics/yoloe-11s
+       ```
+    ## Model Details
+    - **Model**: YOLOE-11S (Small)
+    - **Parameters**: 10.2M
+    - **Format**: PyTorch (.pt)
+    - **Use Case**: Demonstration of official Ultralytics model deployment
+    ## Model Files
+    **Note:** The model weights (`yoloe-11s-seg.pt`) and (`mobileclip_blt.ts`) will be automatically downloaded by ultralytics when the container starts.
+    ## Configuration
+    - **GPU**: Disabled by default (CPU inference)
+    - **Python**: 3.12 with PyTorch 2.3.1+
+    - **Framework**: Ultralytics 8.3+
+    - **Input**: Single image with configurable confidence/IoU thresholds, image size, and class names.
+    - **Output**: Annotated image with detected objects

yoloe11s/cog.yaml

-Original file line number
+Diff line change
@@ -0,0 +1,15 @@
+    # Ultralytics 🚀 AGPL-3.0 License - https://ultralytics.com/license
+    build:
+      gpu: false
+      python_version: "3.12"
+      system_packages:
+        - "libgl1-mesa-glx"
+        - "libglib2.0-0"
+        - "ffmpeg"
+      python_packages:
+        - "ultralytics>=8.3.0"
+        - "clip @ git+https://github.com/ultralytics/CLIP.git@main#egg=clip"
+    predict: predict.py:Predictor
+    image: r8.im/ultralytics/yoloe-11s

yoloe11s/download.py

-Original file line number
+Diff line change
@@ -0,0 +1,21 @@
+    # Ultralytics 🚀 AGPL-3.0 License - https://ultralytics.com/license
+    from pathlib import Path
+    from ultralytics import YOLOE
+    def main():
+        """Download YOLOe-11s-seg and YOLOe-11s-seg-pf weights and move to model directory."""
+        current_dir = Path(__file__).parent
+        YOLOE(current_dir / "yoloe-11s-seg.pt")
+        YOLOE(current_dir / "yoloe-11s-seg-pf.pt")
+        # List files in model directory
+        print(f"Files in {current_dir.name} directory:")
+        for file in sorted(current_dir.iterdir()):
+            print(f"  {file.stat().st_size:>10} {file.name}")
+    if __name__ == "__main__":
+        main()

yoloe11s/predict.py

-Original file line number
+Diff line change
@@ -0,0 +1,54 @@
+    # Ultralytics 🚀 AGPL-3.0 License - https://ultralytics.com/license
+    from typing import Optional
+    from cog import BaseModel, BasePredictor, Input, Path
+    from ultralytics import YOLOE
+    class Output(BaseModel):
+        """Output model for predictions."""
+        image: Optional[Path] = None
+        json_str: Optional[str] = None
+    class Predictor(BasePredictor):
+        """YOLOE: Real-Time Seeing Anything model predictor for Replicate deployment."""
+        def setup(self) -> None:
+            """Load YOLOE-11S model into memory."""
+            self.model = YOLOE("yoloe-11s-seg.pt")
+        def re_init_model(self, class_names: str) -> None:
+            """Re-Initialize model with class names."""
+            if class_names.strip() != "":
+                self.model = YOLOE("yoloe-11s-seg.pt")
+                class_list = class_names.split(", ")
+                self.model.set_classes(class_list, self.model.get_text_pe(class_list))
+            else:
+                # Load YOLOE-11s model prompt free model into memory
+                self.model = YOLOE("yoloe-11s-seg-pf.pt")
+        def predict(
+            self,
+            image: Path = Input(description="Input image"),
+            conf: float = Input(description="Confidence threshold", default=0.25, ge=0.0, le=1.0),
+            iou: float = Input(description="IoU threshold for NMS", default=0.45, ge=0.0, le=1.0),
+            imgsz: int = Input(description="Image size", default=640, choices=[320, 416, 512, 640, 832, 1024, 1280]),
+            class_names: str = Input(
+                description="Comma-separated list of class names to filter results (e.g., 'person, bus, sign') You can also leave it empty to detect classes automatically.",
+                default="person, bus, sign",
+            ),
+            return_json: bool = Input(description="Return detection results as JSON", default=False),
+        ) -> Output:
+            """Run inference and return annotated image with optional JSON results."""
+            self.re_init_model(class_names)
+            result = self.model(str(image), conf=conf, iou=iou, imgsz=imgsz)[0]
+            image_path = "output.png"
+            result.save(image_path)
+            if return_json:
+                return Output(image=Path(image_path), json_str=result.to_json())
+            else:
+                return Output(image=Path(image_path))

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

feat: 🚀 add YOLOE-11S model deployment for replicate #20

Uh oh!

Diff view

Diff view

There are no files selected for viewing

Uh oh!