Sparse Coding on Stereo Video for Object Detection
This work addresses the challenge of data scarcity in computer vision for domains with limited labeled training data, presenting an incremental improvement by integrating sparse coding into existing DCNN frameworks.
The paper tackles the problem of reducing the need for large labeled datasets in deep convolutional neural networks for object detection by using unsupervised sparse coding on stereo video data, showing improved car detection performance with limited labeled examples and more consistent training behavior.
Deep Convolutional Neural Networks (DCNN) require millions of labeled training examples for image classification and object detection tasks, which restrict these models to domains where such datasets are available. In this paper, we explore the use of unsupervised sparse coding applied to stereo-video data to help alleviate the need for large amounts of labeled data. We show that replacing a typical supervised convolutional layer with an unsupervised sparse-coding layer within a DCNN allows for better performance on a car detection task when only a limited number of labeled training examples is available. Furthermore, the network that incorporates sparse coding allows for more consistent performance over varying initializations and ordering of training examples when compared to a fully supervised DCNN. Finally, we compare activations between the unsupervised sparse-coding layer and the supervised convolutional layer, and show that the sparse representation exhibits an encoding that is depth selective, whereas encodings from the convolutional layer do not exhibit such selectivity. These result indicates promise for using unsupervised sparse-coding approaches in real-world computer vision tasks in domains with limited labeled training data.