Modification method for single-stage object detectors that allows to exploit the temporal behaviour of a scene to improve detection accuracy
This addresses the challenge of detecting objects in video data for applications like surveillance or autonomous driving, but it is incremental as it modifies existing methods rather than introducing a new paradigm.
The paper tackles the problem of improving object detection accuracy in videos by modifying single-stage detectors like YOLO and SSD to exploit temporal scene behavior, resulting in considerable accuracy gains, especially for occluded and hidden objects, with a weakly supervised training method that requires no extra annotations.
A simple modification method for single-stage generic object detection neural networks, such as YOLO and SSD, is proposed, which allows for improving the detection accuracy on video data by exploiting the temporal behavior of the scene in the detection pipeline. It is shown that, using this method, the detection accuracy of the base network can be considerably improved, especially for occluded and hidden objects. It is shown that a modified network is more prone to detect hidden objects with more confidence than an unmodified one. A weakly supervised training method is proposed, which allows for training a modified network without requiring any additional annotated data.