-
需要增加SoC型号,CPU、GPU型号在对应的结果bench中展示,在对应的device.py里增加对应方法获取这些信息
-
Is it possible to convert quantized model from other framework?
I see converterOps for quantized tflite but i found only instructions to do QAT in MNN. So should converter work for quantized models?…
-
有谁有办法转成ncnn模型嘛
-
@MaybeShewill-CV Hi.
I am planning to use MNN-LaneNet with VULKAN backend. Is it possible to make this repo with VULKAN backend ? I read cmakelists.txt but there is no any thing about vulkan backend.…
-
## Mobile inference frameworks benchmark (4*ARM_CPU)
Network|VOC mAP(0.5)|COCO mAP(0.5)|Resolution|Inference time (NCNN/Kirin 990)|Inference time (MNN arm82/Kirin 990)|FLOPS|Weight size
:---:|:---:|…
-
Update all pods
Updating local specs repositories
Analyzing dependencies
Downloading dependencies
Installing FMDB (2.7.5)
Installing MNN (0.2.2.2)
[!] Error installing MNN
[!] /usr/bin/curl -…
fatly updated
4 years ago
-
Thank you for the great framework!
I am using mobilenetV2-deeplabV3 on cityscapes converted from [tensorflow](https://github.com/tensorflow/models/blob/master/research/deeplab/g3doc/model_zoo.md). …
-
Whilst the codes in this repository work fine on Kaggle environment (using !pip install for keras_applications, tensorflow-gpu-optimization, mnn, onnxruntime), convolution algorithm fails to load on o…
-
因为一些原因,我想使用静态库的方式使用,研发哥哥能否支持呢。
-
I saw there are code of opengl es as the backend. What is the roadmap for this backend for MNN framework?