removed owlv2

app.py

@@ -70,7 +70,7 @@ async def detect_endpoint(
     video: UploadFile = File(...),
     mode: str = Form(...),
     queries: str = Form(""),
-    detector: str = Form("owlv2_base"),
+    detector: str = Form("hf_yolov8"),
     segmenter: str = Form("sam3"),
 ):
     """
@@ -80,7 +80,7 @@ async def detect_endpoint(
         video: Video file to process
         mode: Detection mode (object_detection, segmentation, drone_detection)
         queries: Comma-separated object classes for object_detection mode
-        detector: Model to use (owlv2_base, hf_yolov8, detr_resnet50, grounding_dino)
+        detector: Model to use (hf_yolov8, detr_resnet50, grounding_dino)
         segmenter: Segmentation model to use (sam3)
 
     Returns:

demo.html

@@ -374,7 +374,6 @@
                 <div class="input-group">
                     <label for="detector">2. Select Detection Model</label>
                     <select id="detector">
-                        <option value="owlv2_base">OWLv2 (Open-vocabulary, Default)</option>
                         <option value="hf_yolov8">YOLOv8 (Fast, COCO classes)</option>
                         <option value="detr_resnet50">DETR ResNet-50 (Transformer-based)</option>
                         <option value="grounding_dino">Grounding DINO (Open-vocabulary)</option>

inference.py

@@ -114,7 +114,7 @@ def run_inference(
         output_video_path: Path to write processed video
         queries: List of object classes to detect (e.g., ["person", "car"])
         max_frames: Optional frame limit for testing
-        detector_name: Detector to use (default: owlv2_base)
+        detector_name: Detector to use (default: hf_yolov8)
 
     Returns:
         Path to processed output video
@@ -133,7 +133,7 @@ def run_inference(
     logging.info("Detection queries: %s", queries)
 
     # Select detector
-    active_detector = detector_name or "owlv2_base"
+    active_detector = detector_name or "hf_yolov8"
     logging.info("Using detector: %s", active_detector)
 
     # Process frames

models/detectors/owlv2.py

@@ -1,56 +0,0 @@
-import logging
-from typing import Sequence
-
-import numpy as np
-import torch
-from transformers import Owlv2ForObjectDetection, Owlv2Processor
-
-from models.detectors.base import DetectionResult, ObjectDetector
-
-
-class Owlv2Detector(ObjectDetector):
-    MODEL_NAME = "google/owlv2-base-patch32"
-
-    def __init__(self) -> None:
-        self.device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
-        logging.info("Loading %s onto %s", self.MODEL_NAME, self.device)
-        self.processor = Owlv2Processor.from_pretrained(self.MODEL_NAME)
-        torch_dtype = torch.float16 if self.device.type == "cuda" else torch.float32
-        self.model = Owlv2ForObjectDetection.from_pretrained(
-            self.MODEL_NAME, torch_dtype=torch_dtype
-        )
-        self.model.to(self.device)
-        self.model.eval()
-        self.name = "owlv2_base"
-
-    def predict(self, frame: np.ndarray, queries: Sequence[str]) -> DetectionResult:
-        inputs = self.processor(text=queries, images=frame, return_tensors="pt")
-        if hasattr(inputs, "to"):
-            inputs = inputs.to(self.device)
-        else:
-            inputs = {
-                key: value.to(self.device) if hasattr(value, "to") else value
-                for key, value in inputs.items()
-            }
-        with torch.no_grad():
-            outputs = self.model(**inputs)
-        processed = self.processor.post_process_object_detection(
-            outputs, threshold=0.3, target_sizes=[frame.shape[:2]]
-        )[0]
-        boxes = processed["boxes"]
-        scores = processed.get("scores", [])
-        labels = processed.get("labels", [])
-        boxes_np = boxes.cpu().numpy() if hasattr(boxes, "cpu") else np.asarray(boxes)
-        if hasattr(scores, "cpu"):
-            scores_seq = scores.cpu().numpy().tolist()
-        elif isinstance(scores, np.ndarray):
-            scores_seq = scores.tolist()
-        else:
-            scores_seq = list(scores)
-        if hasattr(labels, "cpu"):
-            labels_seq = labels.cpu().numpy().tolist()
-        elif isinstance(labels, np.ndarray):
-            labels_seq = labels.tolist()
-        else:
-            labels_seq = list(labels)
-        return DetectionResult(boxes=boxes_np, scores=scores_seq, labels=labels_seq)

models/model_loader.py

@@ -5,13 +5,11 @@ from typing import Callable, Dict, Optional
 from models.detectors.base import ObjectDetector
 from models.detectors.detr import DetrDetector
 from models.detectors.grounding_dino import GroundingDinoDetector
-from models.detectors.owlv2 import Owlv2Detector
 from models.detectors.yolov8 import HuggingFaceYoloV8Detector
 
-DEFAULT_DETECTOR = "owlv2_base"
+DEFAULT_DETECTOR = "hf_yolov8"
 
 _REGISTRY: Dict[str, Callable[[], ObjectDetector]] = {
-    "owlv2_base": Owlv2Detector,
     "hf_yolov8": HuggingFaceYoloV8Detector,
     "detr_resnet50": DetrDetector,
     "grounding_dino": GroundingDinoDetector,