FastDeploy

FastDeploy copied to clipboard

### PR types(PR类型) Benchmark ### Describe Add uie benchmark

joey12300

### PR types(PR类型) Backend ### Describe - Implemented OpenCV_CUDA preprocessors: BGR2RGB, ResizeByShort, CenterCrop, Normalize, NormalizeAndPermute, HWC2CHW - Fuse preprocessors(BGR2RGB/RGB2BGR + Normalize + HWC2CHW). - PPClas integrated OpenCV_CUDA preprocessing - PPClas...

wang-xinyu

### PR types(PR类型) Diffusion ### Describe Add C++ dpm solver

joey12300

### PR types(PR类型) Doc ### Describe - Update PaddleSeg README.md with models exported with argmax op

felixhjh

### PR types(PR类型) Other ### Describe - Update ppseg jni via new api and optimize jni vis func

DefTruth

[Backend] Add Ascend NPU deploy support through PaddleLite CANN

1

1.添加华为昇腾NPU的部署支持. 2.完成分类模型的demo.(测过ResNet50和MobileNetV2) 3.完成相关文档. 4.完成paddledetection（PPYOLOE)/ YOLOv5/v6/v7的支持

yunyaoXYY

ppyoloe模型 多显卡环境使用trt加速，使用第二张显卡出错。Builder was created on device different than current device.

3

## 环境 - FastDeploy版本： 0.7.0 自己编译的 修改了opencv的版本 - 系统平台: Windows x64(Windows10) - 硬件： CUDA 11.2 CUDNN 8.3 N卡2060 12G和2070ti 11G - 编译语言： C++ / Python(3.7或3.8等） ## 问题描述   

yyj2013-two

0.8.0版本ocrv3的predict方法直接程序退出

2

0.8.0版本ocrv3的predict方法直接程序退出，0.7.0版本没有这个问题。测试脚本如下，使用是paddleocr的公开模型，48行执行异常退出，没有报错信息： import os os.environ['CUDA_MODULE_LOADING'] = 'LAZY' import fastdeploy as fd from fastdeploy import ModelFormat import cv2 import numpy as np import os import psutil det_model_file = 'weights/ch_PP-OCRv3_det_infer/inference.pdmodel' det_params_file = 'weights/ch_PP-OCRv3_det_infer/inference.pdiparams'...

yz2yz

[Model] Support YOLOv7-face Model

2

### Describe 1.增加支持Yolov7-face模型 2.该模型来自 https://github.com/derronqi/yolov7-face ，是在YOLOv7基础上对人脸识别进行了优化 ### New feature * Dynamic keypoints * WingLoss * Efficient backbones * EIOU and SIOU | Method | Test Size | Easy | Medium...

CoolKbh

1.detection、face、seg module add "svPreview.enableCamera();". 2.java/android add fastdepolyUi module.

WinterGeng