Estimated Depth Map Helps Image Classification
This work addresses image classification for computer vision practitioners by transferring depth estimation knowledge, but it is incremental as it applies an existing method to a new data combination.
The paper tackles image classification by using estimated depth maps as additional features, showing that this approach improves performance on RGBD datasets derived from RGB ones. It reports accuracy gains of 0.55% for ResNet-20 and 0.53% for ResNet-56.
We consider image classification with estimated depth. This problem falls into the domain of transfer learning, since we are using a model trained on a set of depth images to generate depth maps (additional features) for use in another classification problem using another disjoint set of images. It's challenging as no direct depth information is provided. Though depth estimation has been well studied, none have attempted to aid image classification with estimated depth. Therefore, we present a way of transferring domain knowledge on depth estimation to a separate image classification task over a disjoint set of train, and test data. We build a RGBD dataset based on RGB dataset and do image classification on it. Then evaluation the performance of neural networks on the RGBD dataset compared to the RGB dataset. From our experiments, the benefit is significant with shallow and deep networks. It improves ResNet-20 by 0.55% and ResNet-56 by 0.53%. Our code and dataset are available publicly.