2020 年之后的边缘视觉 SoC

您准备好带着新款智能 SoC 迈向更高水平了吗? 在本视频系列中,将为您带来最新信息,即将视觉智能引入日常使用的嵌入式系统,包括汽车、移动设备、可穿戴设备和相机。各位行业带头人和 Synopsys 的创新者将在此分享他们的专业知识。无论您感兴趣的 AI 领域是极限边缘的深度神经网络、自动驾驶汽车的传感器融合、提早准确估算功耗,还是汽车安全保护功能,都可以获得一些知识,帮助您将创新 SoC 快速推向市场。

Enabling ​DNNs at the Extreme Edge: Co-optimize Circuits, Architectures & Algorithms
DNN inference comes with computational complexity, making execution often feasible only on servers or GPUs. Optimize NPU/TPU processor architectures, dataflow schedulers & quantized NN models for minimum latency & maximum energy efficiency.
Video Player is loading.
Current Time 0:00
Duration 0:00
Loaded: 0%
Stream Type LIVE
Remaining Time 0:00
 
1x
    • Chapters
    • descriptions off, selected
    • subtitles off, selected

        在极限边缘启用 DNN:在电路、架构和算法调度领域协同实现优化

        荷兰天主教鲁汶大学副教授兼比利时微电子研究中心科研总监 Marian Verhelst 

        深度神经网络推理采用极其复杂的计算方式,因而通常只能在能耗高的服务器或 GPU 平台上执行。在边缘和极限边缘设备上进行嵌入式神经网络处理已成为最新趋势,因而必须实现彻底的跨层优化。了解如何优化 NPU/TPU 处理器架构、数据流调度器和量化神经网络模型,以优化延迟和能效。 

        Sensor Fusion for Autonomous Vehicles: Strategies, Methods, and Tradeoffs
        This video presents key sensor fusion strategies for combining heterogeneous sensor data in automotive SoCs. It discusses the three main fusion methods that can be applied in a perception system: early fusion, late fusion and mid-level fusion.
        Video Player is loading.
        Current Time 0:00
        Duration 0:00
        Loaded: 0%
        Stream Type LIVE
        Remaining Time 0:00
         
        1x
          • Chapters
          • descriptions off, selected
          • subtitles off, selected

              自动驾驶汽车传感器融合:策略、方法和权衡

              渥太华大学电气工程和计算机科学学院教授 Robert Laganiere

              了解智能汽车使用的各种传感器的优缺点。了解可用于合并异构传感器数据的主要传感器融合策略,以及感知系统中可以应用的三种主要融合方法:早期融合、晚期融合和中期融合。

              Trends for Embedded Vision & AI IP in Edge Applications
              Discuss the latest trends in machine learning for edge applications, including emerging models like EfficientNet and new techniques like Transformers. Introduce the key challenges and opportunities, with a specific focus on bandwidth optimizatio
              Video Player is loading.
              Current Time 0:00
              Duration 0:00
              Loaded: 0%
              Stream Type LIVE
              Remaining Time 0:00
               
              1x
                • Chapters
                • descriptions off, selected
                • subtitles off, selected

                    边缘应用中嵌入式视觉和 AI 的趋势

                    Synopsys 研发总监 Pierre Paulin

                    由于 1) 计算和内存需求巨大,而且 2) 执行现代视觉和传感任务的算法创新速度不断加快,所以将计算机视觉和深度学习嵌入到边缘仍然面临挑战。敬请了解边缘应用中机器学习的最新趋势,包括 EfficientNet 等新兴模型和转换器等新技术。了解关键的挑战和机遇,并特别关注带宽优化领域。

                    Estimating Power Early & Accurately for Smart Vision SoCs
                    To optimize for both power and performance, HW is tightly intertwined with SW. Understand the key architectural choices such as HW/SW workload partitioning & how/when to estimate power tradeoffs for the most accurate results.
                    Video Player is loading.
                    Current Time 0:00
                    Duration 0:00
                    Loaded: 0%
                    Stream Type LIVE
                    Remaining Time 0:00
                     
                    1x
                      • Chapters
                      • descriptions off, selected
                      • subtitles off, selected

                          提早准确估算智能视觉 SoC 的功耗

                          Synopsys 研发经理 Derya Eker

                          当今的高端 SoC 需要处理日益增长的计算密集型工作负载,但同时必须谨慎权衡功耗与性能。为了优化功耗和性能,软硬件开始更加紧密地结合在一起。了解设计人员在整个开发过程中必须考虑的关键架构选择,例如软硬件工作负载划分,以及何时估算功耗权衡以获得最准确的结果。

                          Safe & Secure SoC Architectures for Autonomous Vehicles
                          Learn about the challenges that SoC designers and OEMs face when developing self-driving vehicles including how a pedestrian looks to software/silicon & understanding an entire scene.
                          Video Player is loading.
                          Current Time 0:00
                          Duration 0:00
                          Loaded: 0%
                          Stream Type LIVE
                          Remaining Time 0:00
                           
                          1x
                            • Chapters
                            • descriptions off, selected
                            • subtitles off, selected

                                适用于自动驾驶汽车的安全保护 SoC 架构

                                Synopsys 研发总监 Fergus Casey

                                先感受软件/硅视角中的行人,再感受整个场景,从而全面了解 SoC 设计人员和 OEM 在开发自动驾驶汽车时面临的挑战。了解在实现自动驾驶的道路上,行业和每个芯片设计必须达成的关键里程碑及其达成标准。