services/detection_service.py

from transformers import DetrImageProcessor, DetrForObjectDetection
import torch
from PIL import Image

class DetectionService:
    def __init__(self, model_name="facebook/detr-resnet-50"):
        self.processor = DetrImageProcessor.from_pretrained(model_name, revision="no_timm")
        self.model = DetrForObjectDetection.from_pretrained(model_name, revision="no_timm")
        self.device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
        self.model.to(self.device)
        self.model.eval()
        self.frame_counter = 0
        self.frame_skip = 5  # Process every 5th frame for performance

    def detect_objects(self, image, confidence_threshold=0.9):
        """Detect objects in an image, skipping frames for performance."""
        self.frame_counter += 1
        if self.frame_counter % self.frame_skip != 0:
            return []  # Skip detection for this frame

        inputs = self.processor(images=image, return_tensors="pt").to(self.device)
        with torch.no_grad():
            outputs = self.model(**inputs)
        target_sizes = torch.tensor([image.size[::-1]]).to(self.device)
        results = self.processor.post_process_object_detection(
            outputs, target_sizes=target_sizes, threshold=confidence_threshold
        )[0]
        detections = []
        for score, label, box in zip(
            results["scores"], results["labels"], results["boxes"]
        ):
            box = box.cpu().numpy().astype(int)
            detections.append({
                "score": score.item(),
                "label": self.model.config.id2label[label.item()],
                "box": {"xmin": box[0], "ymin": box[1], "xmax": box[2], "ymax": box[3]}
            })
        return detections