是否支持rk3568和rk3588?不支持的话是因为硬件原因还是啥原因？

[jinchao123]

No description provided.

[tunmx]

你好RK356X和RK3588的版本还在适配

[jinchao123]

预计什么时候能适配完成？
原始模型哪里可以获取到？想自己转成RK35XX的模型做下适配

[tunmx]

Model files are just archive files, so you can easily get them by adding a .tar extension.

[cuihangmver]

请问支持全志系列的芯片吗，比如全志T507

[charlestamz]

Model files are just archive files, so you can easily get them by adding a .tar extension.

hello, thank you for your response.
I found there's a inspire file in the tar archieve and it should be the model configration file.
could you please give some instructions to write configration ?
for example, how outputs_layers are orgnized?
face_detect:
name: _00_fdet_160
fullname: _00_fdet_160.mnn
version: 0.1
model_type: MNN
infer_engine: MNN
infer_device: MNN
infer_backend: CPU
input_channel: 3
input_image_channel: 3
nchw: true
swap_color: false
data_type: image
input_tensor_type: float32
output_tensor_type: float32
threads: 1
input_layer: input.1
outputs_layers:
- 443
- 468
- 493
- 446
- 471
- 496
- 449
- 474
- 499
input_size:
- 640
- 640
mean:
- 127.5
- 127.5
- 127.5
norm:
- 0.0078125
- 0.0078125
- 0.0078125

[tunmx]

These are the common parameters of the model, such as' outputs_layers' you can open the model file through netron to verify it; Other parameters, such as swap_color, input_channel, mean, etc., are defined based on the actual input parameters required by the model, and you can find them in the source code.

[peijacky]

期待rk3588的版本.......

[aonoa]

Hi, is there any new progress?

[tunmx]

The current version already supports rv1106, will soon support RK356x series, as for RK3588 may have to wait until I purchase the development board to start adapting.

[tunmx]

RK3566/RK3568 Supported! #138