使用神经处理器实现持续感应的摄像头.pdf

1、Using a Neural Processor for Always-Sensing CamerasSharad CholeChief Technologist,Co-founderExpedera Like always-listening Siri,always-sensing enables a more natural and seamless user experience Quality and richness of camera data require much more processing than audio implementations Like always-l

2、istening,always-sensing must be processed locally Technical requirements:500 GOPS to 1 TOPS NPU Ultra-low power Ultra-small area Multiple modelsAlways-sensing Cameras:What and Why2 2023 Expedera Inc.Optimized edge AI inference IP solutions based on revolutionary packet architecture,application-confi

3、gured for our customersSilicon Valley startup founded in 2018 3 R&D centers,numerous patentsBroad,worldwide deployments 10M+devices in-field Over 200 ExaOps (200,000,000,000,000,000,000 operations/second)deployed by our customers Soft IP:designs in multiple leading-edge nodesAbout Expedera3 2023 Exp

4、edera Inc.Like always-listening Siri,always-sensing enables a more natural and seamless user experience Quality and richness of camera data require much more processing than audio implementations Like always-listening,always-sensing must be processed locally Technical requirements:500 GOPS to 1 TOPS

5、 NPU Ultra-low power Ultra-small area Multiple modelsAlways-sensing Cameras:What and Why4 2023 Expedera Inc.NPUs are used in wake-word(audio)applications-why cant the same approach apply to video?Workloads Edge-friendly vision NNs ISPs/DSPs arent designed for this sort of specialized processingWe as

6、ked ourselves rather,customers asked us can we apply the same low power,small,targeted workload approach to video?We have;otherwise,I wouldnt be giving this talkNPUs and Always-Sensing Are an Ideal Match 2023 Expedera Inc.024Yolo V5nMobileNet V3sNanoDet+024Yolo V5nMobileNet V3sNanoDet+01530Yolo V5nM