Where is source code for rknn-toolkit2?

hlacik · March 12, 2023, 1:03am

I mean they have their repo https://github.com/rockchip-linux/rknn-toolkit2
where you can download their compiled wheel packages for ubuntu (https://github.com/rockchip-linux/rknn-toolkit2/blob/master/packages/rknn_toolkit2-1.4.0_22dcfef4-cp36-cp36m-linux_x86_64.whl) as well as for rknn_toolkit_lite2 , but where is the source code to compile it yourself … or is it closed source??

hlacik · March 12, 2023, 10:54pm

I am looking exactly for source code of their rknn-toolkit2 , in other word source code of the rknn_toolkit2.whl package
https://github.com/rockchip-linux/rknn-toolkit2/blob/master/packages/rknn_toolkit2-1.4.0_22dcfef4-cp36-cp36m-linux_x86_64.whl

as well as their runtime library - rknn.so

hlacik · March 12, 2023, 10:55pm

@JonGroff no its not
just pre-built whl packages and example files without source code of their rknn libraries!

hlacik · March 12, 2023, 10:58pm

ok so this example

github.com

rockchip-linux/rknn-toolkit2/blob/324671aa65785fd471920771de98baf9ca45ab43/examples/pytorch/resnet18/test.py#L3


import numpy as np
import cv2
from rknn.api import RKNN
import torchvision.models as models
import torch
import os




def export_pytorch_model():
    net = models.resnet18(pretrained=True)
    net.eval()
    trace_model = torch.jit.trace(net, torch.Tensor(1, 3, 224, 224))
    trace_model.save('./resnet18.pt')

does

from rknn.api import RKNN

i.e importing their RKNN library

I am asking where is the source code of that library?

github repo rknn-toolkit2 contains just prebuilt python libraries and examples

hlacik · March 12, 2023, 10:59pm

thats what i am asking for, so why is it closed source? as far as I have seen – everything for rockchip rk3588 is released as opensource, and this is their only closed source part – why?

hlacik · March 12, 2023, 11:08pm

so than they are same EVIL as NVIDIA with their CUDA.

Yes i am successfully using their python binding (rknn-toolkit-lite2) for inference of our models.

But i am doing proof-of-concept for our next generation of product, which should have long therm support (5+ years). And rn. we have switched from nvidia jetson platform because of their closed source approach to everything.
So i hit the wall with rockchip as well, and when they decide to give up on it in 1 year, oh well – we are doomed source code would def. help to troubleshoot it on own if that happens, or to give it a community – to take care

Thank you very much for your response, appreciate it tho.

hlacik · March 12, 2023, 11:13pm

I’ll ask radxa guys on discord forum, if they have anything to say …

Daryl_Thomas · April 23, 2023, 12:37pm

Hi I am also looking for the source code to build rknn-toolkit2 on my own SBC aarch64, the wheels are only for ubuntu. The toolkit-lite I have installed and run only has limited features. I want to try and run YOLOV5 and YOLOV8 on the R3588S board in Orange PI 5. Is there pre-built wheels for rknn-toolkit available anywhere for aarch64?

hlacik · April 23, 2023, 8:00pm

hey, rknn-toolkit2 is only for PC (x64) and its purpose is to convert your model (from onnx,pytorch,tensorfow,tflite) to RKNN format. rknn-toolkit2 is not for inference.

to run model on arm64 you must use rknn-toolkit-lite-2 which is for RUNNING (inference) on your SBC.

All is here — https://github.com/rockchip-linux/rknn-toolkit2

As to source code – unfortunatelly – it is closed source. hope guys from rockchip will consider opensourcing it in future

Daryl_Thomas · April 28, 2023, 7:57am

Hey yes, I implemented the sample in the rknn-toolit2 github. I managed to run their sample yolov5 through the toolkit-lite-2 framework. I want to now try and run on my own custom models, but the toolkit is using older versions of every library, so that seems to be challenging to keep the code compatible with the other libraries like yolo which are moving at a way higher pace.

hlacik · May 2, 2023, 10:59am

yes, i am struggling with it as well, i hope guys from rockchip will release newer version.