ncnn is a high-performance neural network inference framework optimized for the mobile platform
-
Updated
Oct 25, 2021 - C
{{ message }}
ncnn is a high-performance neural network inference framework optimized for the mobile platform
Video, Image and GIF upscale/enlarge(Super-Resolution) and Video frame interpolation. Achieved with Waifu2x, Real-ESRGAN, SRMD, RealSR, Anime4K, RIFE, CAIN, DAIN, and ACNet.
A lossless video/GIF/image upscaler achieved with waifu2x, Anime4K, SRMD and RealSR. Started in Hack the Valley 2, 2018.
TNN: developed by Tencent Youtu Lab and Guangying Lab, a uniform deep learning inference framework for mobile、desktop and server. TNN is distinguished by several outstanding features, including its cross-platform capability, high performance, model compression and code pruning. Based on ncnn and Rapidnet, TNN further strengthens the support and performance optimization for mobile devices, and also draws on the advantages of good extensibility and high performance from existed open source efforts. TNN has been deployed in multiple Apps from Tencent, such as Mobile QQ, Weishi, Pitu, etc. Contributions are welcome to work in collaborative with us and make TNN a better framework.
Deep Learning API and Server in C++14 support for Caffe, Caffe2, PyTorch,TensorRT, Dlib, NCNN, Tensorflow, XGBoost and TSNE
MobileNetV2-YoloV3-Nano: 0.5BFlops 3MB HUAWEI P40: 6ms/img, YoloFace-500k:0.1Bflops 420KB
AoE (AI on Edge,终端智能,边缘计算) 是一个终端侧AI集成运行时环境 (IRE),帮助开发者提升效率。
is:issue is:open 请问,方便提供一下onnx文件导出的代码文件吗?用model文件夹里面跟随的文件输出结果用ncnn推理结果不对
Generate a quantization parameter file for ncnn framework int8 inference
PFLD pytorch Implementation
mobilev2-yolov5s剪枝、蒸馏,支持ncnn,tensorRT部署。ultra-light but better performence!
全平台实时人脸检测和姿态估计,提供无需任何框架实现Realtime Face Detection and Head pose estimation on Windows、Ubuntu、Mac、Android and iOS
Lite.AI.ToolKit
PyTorch Project Specification.
DAIN, Depth-Aware Video Frame Interpolation implemented with ncnn library
The outputs are great, although not the same as yolov5, maybe some pre-processing/post-processing steps are different.
That's a great catch! I think it is caused by the different pre-processing operations. I've verified the the post-processing stages before, it can get the same results as ultralytics/yolov5 (when w/o TTA predict). And I've uploaded a [notebook](https://github.com/zhiqwang/yol
基于insightface训练mobilefacenet的相关步骤及ncnn转换流程
Add a description, image, and links to the ncnn topic page so that developers can more easily learn about it.
To associate your repository with the ncnn topic, visit your repo's landing page and select "manage topics."
我发现很多人在使用voc格式的数据集时,和我遇到了同样的问题,训练时AP一直为0,
今早,仔细检查后,我也找到了真正的原因,主要是数据加载的地方出现了问题,还是我们自己太不仔细了
解决流程思路: 解决YOLOX训练时AP为0