Which Object Detection architecture (be it CNN-based or Visual Transformer-based) can be used to detect small objects ?
Here are some object detection models that can identify small objects ¹ ²:
- Feature-Fused SSD: This model focuses on sustaining features of small objects passing through convolutional networks.
- YOLO-Z: This model also focuses on sustaining features of small objects passing through convolutional networks.
- Feature Pyramid Network (FPN): This model uses a feature pyramid to learn features on a multi-scale. Examples include Twin Feature Pyramid Networks (TFPN) and Extended Feature Pyramid Network (EFPN).
- YOLOv5: This model uses an auto-anchor algorithm to find good anchors based on the nature of object sizes in the dataset.
- yolov8s-1280: This model is trained on higher-resolution images, making it ideal for detecting small objects.
- Inference Slicer: This model splits the scene into smaller parts, runs the model on each part, and then stitches the results together.
- Badges
- Science topic
- Similar topics
- Academic Journals