Object detection is an essential component of many systems used, for example, in advanced driver assistance systems (ADAS) or advanced video surveillance systems (AVSS). Currently, the highest detection accuracy is achieved by solutions using deep convolutional neural networks (DCNN). Unfortunately, these come at the cost of a high computational complexity; hence, the work on the widely understood acceleration of these algorithms is very important and timely. In this work, we compare three different DCNN hardware accelerator implementation methods: coarse-grained (a custom accelerator called LittleNet), fine-grained (FINN) and sequential (Vitis AI). We evaluate the approaches in terms of object detection accuracy, throughput and energy usage on the VOT and VTB datasets. We also present the limitations of each of the methods considered. We describe the whole process of DNNs implementation, including architecture design, training, quantisation and hardware implementation. We used two custom DNN architectures to obtain a higher accuracy, higher throughput and lower energy consumption. The first was implemented in SystemVerilog and the second with the FINN tool from AMD Xilinx. Next, both approaches were compared with the Vitis AI tool from AMD Xilinx. The final implementations were tested on the Avnet Ultra96-V2 development board with the Zynq UltraScale+ MPSoC ZCU3EG device. For two different DNNs architectures, we achieved a throughput of 196 fps for our custom accelerator and 111 fps for FINN. The same networks implemented with Vitis AI achieved 123.3 fps and 53.3 fps, respectively.

Testing and evaluation of an automotive perception system is a complicated task which requires special equipment and infrastructure. To compute key performance indicators and compare the results with real-world situation, some additional sensors and manual data labelling are often required. In this article, we propose a different approach, which is based on a UAV equipped with a 4K camera flying above a test track. Two computer vision methods are used to precisely determine the positions of the objects around the car – one based on ArUco markers and the other on a DCNN (we provide the algorithms used on GitHub). The detections are then correlated with the perception system readings. For the static and dynamic experiments, the differences between various systems are mostly below 0.5 m. The results of the experiments performed indicate that this approach could be an interesting alternative to existing evaluation solutions.

While neural network quantization effectively reduces the cost of matrix multiplications, aggressive quantization can expose non-matrix-multiply operations as significant performance and resource bottlenecks on embedded systems. Addressing such bottlenecks requires a comprehensive approach to tailoring the precision across operations in the inference computation. To this end, we introduce scaled-integer range analysis ( SIRA ), a static analysis technique employing interval arithmetic to determine the range, scale, and bias for tensors in quantized neural networks. We show how this information can be exploited to reduce the resource footprint of FPGA dataflow neural network accelerators via tailored bitwidth adaptation for accumulators and downstream operations, aggregation of scales and biases, and conversion of consecutive elementwise operations to thresholding operations. We integrate SIRA -driven optimizations into the open source FINN framework, then evaluate their effectiveness across a range of quantized neural network workloads and compare implementation alternatives for non-matrix-multiply operations. We demonstrate an average reduction of 17% for LUTs, 66% for DSPs, and 22% for accumulator bitwidths with SIRA optimizations, providing detailed benchmark analysis and analytical models to guide the implementation style for non-matrix layers. Finally, we open source SIRA to facilitate community exploration of its benefits across various applications and hardware platforms.

Testing and evaluation of an automotive perception system is a complicated task which requires special equipment and infrastructure. To compute key performance indicators and compare the results with the real-world situation, some additional sensors and manual data labelling are often required. In this article, we propose a different approach, which is based on a UAV equipped with a 4K camera flying above the test track. Thanks to the synchronisation of the sensors between the tested vehicle and the UAV, it is possible to precisely determine the positions of the objects around the car and correlate them with the perception system readings. The performed experiments indicate that this approach could be an interesting alternative to the existing evaluation solutions.