We benchmark several widely-used deep learning frameworks and investigate the FPGA deployment for performing traffic sign classification and detection. We evaluate the training speed and inference accuracy of these frameworks on the GPU by training FPGA-deployment-suitable models with various input sizes on GTSRB, a traffic sign classification dataset. Then, selected trained classification models and various object detection models that we train on GTSRB's detection counterpart (i.e., GTSDB) are evaluated with inference speed, accuracy, and FPGA power efficiency by varying different parameters such as floating-point precisions, batch sizes, etc. We discover that Neon and MXNet deliver the best training speed and classification accuracy on the GPU in general for all test cases, while TensorFlow is always among the frameworks with the highest inference accuracies. We observe that with the current OpenVINO release, the performance of lightweight models (e.g., MobileNet-v1-SSD, etc) usually exceeds the requirement of real-time detection without losing much accuracy, while other models (e.g., VGG-SSD, ResNet-50-SSD) generally fail to do so. We also demonstrate that we can adjust the precision of bitstreams and the batch sizes to balance inference speed and accuracy of the applications deployed on the FPGA. Finally, we show that for all test cases, the FPGA always achieves higher power efficiency than the GPU.

We implement a speed-limit-sign recognition task using a template-based approach on the FPGA using the Intel FPGA SDK for OpenCL. Then we evaluate its throughput, power consumption, accuracy, and development effort against a GPU implementation that is based on a system presented in our previous study. This paper also discusses implementation differences between the FPGA and GPU systems, provides a methodology for translating the GPU system to the FPGA system, and explains optimizations used in the FPGA version. While implementing the FPGA system, we build an efficient FFT engine for image processing on the FPGA which can be utilized by other developers to perform related tasks. In this paper, we also provide our insights on building the FPGA versus GPU system, which we hope can be useful for designing upcoming versions of FPGA-focused OpenCL development environments. We conclude that the FPGA implementation provides better power consumption for the same detection accuracy, while the GPU supports better programmer efficiency.