Modified Cascade RCNN Based on Contextual Information for Vehicle Detection

In the process of traditional vehicle detection, there are some problems such as the fault detection and missing detection for small objects and shielded objects. Therefore, we propose a modified Cascade region-based convolutional neural network (RCNN) based on contextual information for vehicle detection. Firstly, the feature pyramid is improved to integrate the shallow information into the deep network layer by layer to enhance the features of small objects and occlusion objects. In here, we introduce the predictive optimization module and combine the context information of the region of interest (ROI), which makes the feature information have stronger robustness. Meanwhile, the multi-scale and multi-stage prediction is realized through the multi-threshold prediction network of internal cascade. Under the premise that the network parameters are basically unchanged, the accuracy rate is improved. Secondly, the multi-branch dilated convolution is introduced to reduce the feature loss during the down-sampling process. Finally, the region of interest and context information are fused to enhance the object feature expression. Experimental results show that the new Cascade RCNN method can better detect small and shielded vehicles compared with other state-of-the-art vehicle detection methods.

Automated summary

The modified Cascade RCNN method proposed in this article utilizes improved feature pyramid, predictive optimization module, and multi-scale prediction network to enhance accuracy in vehicle detection. Experimental results demonstrate its superior performance in small and shielded vehicle detection compared to other state-of-the-art methods.