Computer Vision — Object Detection, One-Stage vs Two-Stage detectors

• What is Object Detection
• Difference between One-Stage and Two-Stages detectors
• Applications of One-Stage and Two-Stages detectors

What is Object Detection

In Computer vision, object detection simply means identifying and locating objects within an image. Unlike image classification, which assigns a single label to an image, object detection uses machine learning algorithms to detect multiple objects and their respective locations within an image. It typically involves two tasks:

1. Object Classification: Identifying what types of objects are present in an image (e.g., cars, people, animals).
2. Object Localization: Determining where these objects are located within the image by drawing bounding boxes around them.

Below is the overview of Object Detection’s history

Zou et al. 2019. Object Detection in 20 Years: A Survey

Difference between one-stage and two-stage

By SharkYun

After 2014, the object detection generally split into 2 categories: one-stage and two-stage detectors. The main difference between these two is the region proposals, where two-stage detectors do generate region proposals and one-stage detectors do not.

Two-stage detectors

• Stage 1: Region proposals generation, typically using selective search or the Region Proposal Network (RPN).
• Stage 2: Object classification and bounding box refinement on each proposed region.

Two-stage approach allows for high accuracy, as the second stage refines results, but it’s generally slower because of the two-step process. The classic two-stage methods is the R-CNN (Region-based Convolutional Neural Networks) series, including Fast R-CNN and Faster R-CNN.

Below is an example of two-stages detector: R-CNN, which it extract the region proposals from the image first.

R-CNN

One-stage detectors

One-stage detectors process the input image by skipping the proposal generation and directly detecting objects in a single pass through the image. The YOLO (You Only Look Once) and SSD (Single Shot MultiBox Detector) models are prime examples of one-stage detectors.

• Single Stage: Divides the image into a grid and simultaneously predicts bounding boxes and class scores for each region.

One-stage detectors are typically faster and more suitable for real-time applications, but they may sacrifice some accuracy compared to two-stage methods.

Below is an example of one-stages detector: YOLO, which it divides an image into a grid, predicts bounding boxes with confidence scores, and uses non-maximum suppression (NMS) to finalize object detections.

YOLO

Application

Two-stage detectors: Known for higher accuracy, especially in complex scenes, due to their ability to refine proposals. Faster R-CNN, for instance, is often preferred in applications where precision is crucial.

• Use case: Medical image analysis, facial recognition, satellite imagery, and other fields where high detection accuracy is essential over speed.

One-stage detectors: Known for their speed, making them ideal for applications where latency is the key factor.

• Use case: Real-time detection like autonomous vehicles and surveillance cameras

Conclusion

Even though two-stage detectors are more accurate due to their structure, advancements in one-stage detectors have significantly narrowed the performance gap. (As of Oct 2024, YOLO v11 just released.)

End of the day, the choice between one-stage and two-stage detectors depends more on specific application needs rather than a clear trade-off in performance.

Reference:

https://www.researchgate.net/publication/353284602_Semantic_Image_Cropping

Object Detection in 20 Years: A Survey

What is R-CNN?

You Only Look Once: Unified, Real-Time Object Detection