NVIDIA formally released JetPack-3.2 around mid March. I finally got time to update my Jetson TX2 to this latest BSP release and started to play with it.

I tested/verified my previous posts with JetPack-3.2. Overall I only needed to do some minor updates to my installation steps for opencv3. All other stuffs worked the same way as before (JetPack-3.1). More specifically, I’ve verified (and upated if necessary) the following:

• How to Install OpenCV (3.4.0) on Jetson TX2
• How to Install Caffe and PyCaffe on Jetson TX2
• How to Capture and Display Camera Video with Python on Jetson TX2
• How to Capture Camera Video and Do Caffe Inferencing with Python on Jetson TX2
• YOLOv2 on Jetson TX2
• Single Shot MultiBox Detector (SSD) on Jetson TX2
• Faster R-CNN on Jetson TX2
• YOLOv3 on Jetson TX2

In addition to the above, I think the most significant update in JetPack-3.2 is the release of TensorRT 3.0 GA. Reference: JetPack Release Notes. I shall look into TensorRT 3.0 in the near future. Hopefully I could share some experiences on that soon.

TensorRT 3.0 GA
  - Support for TensorFlow models
  - Up to 15% perf/W improvement for DL applications