Nvidia JetSon Nano Developer Kit : Not able to Install or not run able to run EfficientViT

Robotics & Edge Computing Jetson & Embedded Systems Jetson Nano
1

Source # Tutorial - EfficientViT EfficientViT - NVIDIA Jetson AI Lab

  1. After running jetson-containers run $(autotag efficientvit) it does not completely built the container image to run efficientvit. Why am i raising this as an issue it took couple of hours to try this and after multiple tries to get the container up and running

/opt directory was missing
i created the directory in the container image

mkdir /opt/efficientvit
  1. the benchmark.py was missing and copied that to
docker cp benchmark.py 5395cc7a829f:/opt/efficientvit/benchmark.py
  1. The effecientvit_12.pt location was change and downloaded into the nvidia nano data directory data/models/efficientvit/sam./12.pt

Soruce:
efficientvit_sam_l2.pt · mit-han-lab/efficientvit-sam at main
efficientvit_sam_l2.pt · mit-han-lab/efficientvit-sam at main

  1. Then it was about trying to deploy the model manually

python3 benchmark.py 
Traceback (most recent call last):
  File "/opt/efficientvit/benchmark.py", line 10, in <module>
    import matplotlib.pyplot as plt
ModuleNotFoundError: No module named 'matplotlib'
root@nvidiaai:/opt/efficientvit# pip install matplotlib

python3 benchmark.py 
Traceback (most recent call last):
  File "/opt/efficientvit/benchmark.py", line 16, in <module>
    from efficientvit.apps.utils import parse_unknown_args
ModuleNotFoundError: No module named 'efficientvit'

Tried to build it manually but their were other issues

Note : I think there are other uses who faced similar issue as well

1 Like
3

Hi,

Suppose you are using JetPack 6.2, is that correct?

We didn’t update efficientvit for a while so some modification should be needed to make it work on JetPack 6.2.
For example, the Dockerfile still uses python 3.8 branch but the default Python for r36.4.3 is 3.10 already.

Thanks.

1 Like
4

yes indeed i am on JetPack 6.2

5

Hi,

So you will need to make some changes on the Dockerfile to make it work on JetPack 6.2.
Thanks.

6

Please can you let me know and i am happy it test it

7

Hi,

We don’t have a concrete schedule to update the container for the latest release.
But as these sample are all open sourced, it’s more recommended to try it directly.

Thanks.

1 Like
8

The reason i started logging this or other issues was that most of the samples did not work on the latest hardware

11

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.