On Exploring Image Resizing for Optimizing Criticality-based Machine Perception
On Exploring Image Resizing for Optimizing Criticality-based Machine Perception
复制标题
探索图像大小调整以优化基于关键性的机器感知
DOI：
10.1109/rtcsa52859.2021.00027
复制
发表时间：
2021
期刊：
影响因子：
--
通讯作者：
P. David
中科院分区：
文献类型：
--
作者：
Yigong Hu;Shengzhong Liu;T. Abdelzaher;Maggie B. Wigness;P. David
在诸如自动驾驶无人机或汽车等嵌入式硬件上运行的现代机器推理流水线中，板载计算能力仍然是一个关键瓶颈。为了缓解这一瓶颈，近期的研究提出了一种架构，用于对复杂模态（如视频）的输入帧进行分割，并根据感知场景各部分的关键性对下游机器感知任务进行优先级排序。基于关键性的优先级排序使得有限的机器资源（低端嵌入式GPU）能够更明智地首先用于跟踪更重要的对象。本文探索了基于关键性的机器感知优先级排序的一个新维度；即，依赖关键性的图像调整大小作为一种改善感知质量和及时性之间权衡的方法所起的作用。给定关键性评估（例如，物体与自动驾驶汽车的距离），调度器在将调整大小后的图像传递给感知模块之前，可以从几种图像调整大小选项（以及相关的推理模型）中进行选择。在一个具有真实驾驶数据集的人工智能嵌入式平台上进行的实验表明，当使用所提出的调整大小算法时，在感知准确性和响应时间之间的权衡上有显著改善。这种改善归因于所提出方案的两个优势：(i) 通过减少在不太关键的对象上花费的时间，改善了对更关键对象的优先处理，以及 (ii) 由于重新调整大小，改善了GPU内的图像批处理，从而提高了资源利用率。
On-board computing capacity remains a key bottleneck in modern machine inference pipelines that run on embedded hardware, such as aboard autonomous drones or cars. To mitigate this bottleneck, recent work proposed an architecture for segmenting input frames of complex modalities, such as video, and prioritizing downstream machine perception tasks based on criticality of the respective segments of the perceived scene. Criticality-based prioritization allows limited machine resources (of lower-end embedded GPUs) to be spent more judiciously on tracking more important objects first. This paper explores a novel dimension in criticality-based prioritization of machine perception; namely, the role of criticality-dependent image resizing as a way to improve the trade-off between perception quality and timeliness. Given an assessment of criticality (e.g., an object’s distance from the autonomous car), the scheduler is allowed to choose from several image resizing options (and related inference models) before passing the resized images to the perception module. Experiments on an AI-powered embedded platform with a real-world driving dataset demonstrate significant improvements in the trade-off between perception accuracy and response time when the proposed resizing algorithm is used. The improvement is attributed to two advantages of the proposed scheme: (i) improved preferential treatment of more critical objects by reducing time spent on less critical ones, and (ii) improved image batching within the GPU, thanks to re-sizing, leading to better resource utilization.