Merge branch 'main' into codeflash/optimize-select_top_confidence_detection-m7drsd4o

Author: PawelPeczek-Roboflow

docs/fine-tuned/rfdetr.md

@@ -0,0 +1,65 @@
+RF-DETR is Roboflow's new Real-Time SOTA Object Detector.
+
+## Supported Model Types
+
+You can deploy the following RF-DETR model types with Inference:
+
+- Object Detection
+
+## Model Overview
+
+- [Read More About RF-DETR](https://blog.roboflow.com/rf-detr/)
+
+## Usage Example
+
+```
+import os
+import supervision as sv
+from inference import get_model
+from PIL import Image
+from io import BytesIO
+import requests
+
+response = requests.get("https://media.roboflow.com/dog.jpeg")
+
+if response.status_code == 200:
+    image_data = BytesIO(response.content)
+
+    image = Image.open(image_data)
+
+model = get_model("rfdetr-base")
+
+predictions = model.infer(image, confidence=0.5)[0]
+
+detections = sv.Detections.from_inference(predictions)
+
+labels = [prediction.class_name for prediction in predictions.predictions]
+
+annotated_image = image.copy()
+annotated_image = sv.BoxAnnotator().annotate(annotated_image, detections)
+annotated_image = sv.LabelAnnotator().annotate(annotated_image, detections, labels)
+
+sv.plot_image(annotated_image)
+
+annotated_image.save("annotated_image_base.jpg")
+
+model = get_model("rfdetr-large")
+
+predictions = model.infer(image, confidence=0.5)[0]
+
+detections = sv.Detections.from_inference(predictions)
+
+labels = [prediction.class_name for prediction in predictions.predictions]
+
+annotated_image = image.copy()
+annotated_image = sv.BoxAnnotator().annotate(annotated_image, detections)
+annotated_image = sv.LabelAnnotator().annotate(annotated_image, detections, labels)
+
+sv.plot_image(annotated_image)
+
+annotated_image.save("annotated_image_large.jpg")
+```
+
+## License
+
+See our [Licensing Guide](https://roboflow.com/licensing) for more information about how your use of RF-DETR is licensed when using Inference to deploy your model.

inference/core/roboflow_api.py

@@ -4,6 +4,7 @@
 import re
 import urllib.parse
 from enum import Enum
+from hashlib import sha256
 from pathlib import Path
 from typing import Any, Callable, Dict, List, Optional, Tuple, Type, Union
 
@@ -555,8 +556,9 @@ def get_workflow_specification(
         if not re.match(r"^[\w\-]+$", workflow_id):
             raise ValueError("Invalid workflow id")
 
+        workflow_hash = sha256(workflow_id.encode()).hexdigest()
         local_file_path = (
-            Path(MODEL_CACHE_DIR) / "workflow" / "local" / f"{workflow_id}.json"
+            Path(MODEL_CACHE_DIR) / "workflow" / "local" / f"{workflow_hash}.json"
         )
         if not local_file_path.exists():
             raise FileNotFoundError(f"Local workflow file not found: {local_file_path}")

inference/core/version.py

@@ -1,4 +1,4 @@
-__version__ = "0.43.0"
+__version__ = "0.44.0"
 
 
 if __name__ == "__main__":

inference/core/workflows/core_steps/common/query_language/operations/detections/base.py

@@ -113,19 +113,19 @@ def select_top_confidence_detection(detections: sv.Detections) -> sv.Detections:
 
 def select_leftmost_detection(detections: sv.Detections) -> sv.Detections:
     if len(detections) == 0:
-        return deepcopy(detections)
+        return detections  # Directly return the original empty detections if empty
+
     centers_x = detections.get_anchors_coordinates(anchor=Position.CENTER)[:, 0]
-    min_value = centers_x.min()
-    index = np.argwhere(centers_x == min_value)[0].item()
+    index = np.argmin(centers_x)
     return detections[index]
 
 
 def select_rightmost_detection(detections: sv.Detections) -> sv.Detections:
     if len(detections) == 0:
-        return deepcopy(detections)
+        return detections
+
     centers_x = detections.get_anchors_coordinates(anchor=Position.CENTER)[:, 0]
-    max_value = centers_x.max()
-    index = np.argwhere(centers_x == max_value)[-1].item()
+    index = centers_x.argmax()
     return detections[index]