-
Notifications
You must be signed in to change notification settings - Fork 473
New issue
Have a question about this project? # for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “#”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? # to your account
FastDeploy模型、硬件支持计划 #6
Comments
这个只是针对图像类的模型吗? |
@datalee 可以,目前已经有UIE。 PaddleNLP的所有部署之后都会由FastDeploy来支持哈 |
#369 PP-ShiTu模型 |
|
#358 3D目标检测模型 |
|
mnn 或ncnn 计划支持么,非常期待 |
#373 PPTracking已经支持 |
有对比过和onnx之间的性能差异吗? |
请问大佬有对tracking模型和像pphuman 这样的action识别流程的支持计划吗 |
#465 YOLOv5-seg |
有没有支持windows 32? |
如果有外部开发者有兴趣来做这个事的话,我可以帮忙一起Review和说明如何集成 |
Tracking已经集成支持 |
现在都基本是Windows x64了呢,你在什么场景下还依赖windows 32呢 |
paddlex 训练的模型能通过这个部署嘛 |
正在支持中 |
paddleX 期待 |
paddleX 期待 +1 |
硬件型号:Nvidia T4 需求描述:希望FastDeploy支持nvidia gpu的硬解和推流 |
硬件型号:RK3588 |
|
@populustremble FastDeploy Streamer已经有相应的支持,可参考https://github.com/PaddlePaddle/FastDeploy/tree/develop/streamer
|
这部分功能有计划实现python的api吗 |
@populustremble 有计划,预计3月底之前会有python API |
硬件型号:Nvidia Jetson nano b01 4G 需求描述:希望FastDeploy支持在nvidia jetson nano上面进行脸部检测的trt部署,现在根据我测试的结果 yolov7face retinaface 模型还不能在nano上面实现trt加速预测功能 |
硬件型号:Nvidia T4 需求描述:strongsort、ocsort、bytetrack、botsort 模型支持 |
硬件型号:nvidia-smi A40 和A100 |
硬件型号:昇腾310 |
KIE模型有支持计划么 |
GTX750Ti |
有没有文档矫正的功能 |
请问 我们需要PPHGNet samll版 分类模型 的rknn模型转换支持,尝试使用现在公开的 ResNet50_vd |
NLP模型支持的好少,未来会支持更多的大模型吗?不只是PaddleNLP的。 |
支持GPT2 #2051 |
BM1684期待有更好的支持,现在使用examples目录下的YOLOV5的cpp推理,会卡住不动,下面是执行日志。 |
YOLOv8 |
硬件型号:Intel J1900、Celeon系列 实测发现paddleserving部署OCR,因为无AVX指令因此失败。 请问FastDeploy是否支持无AVX的intel CPU,若不支持,是否有支持计划。 |
目前需要avx指令的应该是Paddle Inference后端。可以自行编译FastDeploy,仅集成ONNXRuntime推理后端即可。 |
swin transformer模型是否有计划支持 |
支持paddlespeech吗 |
PP-YOLOE-R 旋转框,是否能支持 |
硬件型号:RTX4070 |
硬件型号:RK3588 |
模型 (CV、NLP、Speech、Cross-modal)
平台& 硬件 (云、边、端)
如果你希望FastDeploy支持新的AI模型/新硬件,或与FastDeploy共建生态,欢迎提供如下信息我们:
The text was updated successfully, but these errors were encountered: