superbayes

有没有预训练的模型?

1

作者大佬,能否提供您基于人体姿态的模型,作为预训练的基础模型? 这样当我们训练自己特定的数据,比如手部关键点识别,就可以在预训练的模型上再次训练,加快训练速度. thanks a lot!!!

yolov5n.pt 模型转ncnn后,预测结果为空

6

## detail | 详细描述 | 詳細な説明 ## 我最近使用yolov5r6官方提供的yolov5n.pt,转换为ncnn模型后,无法预测出物体 我的步骤是: 1: `python export.py --weights models/yolov5n.pt --include onnx --train` 2: `python -m onnxsim models/yolov5n.onnx models/yolov5n-sim.onnx` 3: 在 [转换模型网址](https://convertmodel.com/) 将模型转换为 yolov5n-sim-opt-fp16 模型 4,使用yolov5进行预测,但没有结果...

could you offer more details about the model inference？

1

1，please offer the cpu type which you use; 2, please offer the model size

ld: error: found local symbol '__bss_start' in global part of symbol table in file

1

当我构建项目到手机时,发生错误.报错日志如下:  **ld: error: found local symbol '__bss_start' in global part of symbol table in file F:/MyCode/androidStudio/NDK_LearnOpengl30/NDK_OpenGLES_3_0/app/src/main/cpp/../jniLibs/arm64-v8a/libopencv_java3.so** ### 我该如何处理?

如果input size=640*640,那么MAP可能会进一步提高吗?

1

如果nanodet支持input size=640*640(像YOLOV5一样),那么可能MAP比同级别的yoloV5高,并且推断速度也比yoloV5快. 这种可能性高吗? 实际应用场景应该有很多吧?

# old : * 1, you need Generate a dataset path .txt file ``` train.txt /home/qiuqiu/Desktop/dataset/train/000001.jpg /home/qiuqiu/Desktop/dataset/train/000002.jpg val.txt /home/qiuqiu/Desktop/dataset/val/000070.jpg /home/qiuqiu/Desktop/dataset/val/000043.jpg ``` * 2, your coco.yaml like this ``` DATASET: TRAIN:...

I have discovered a C++source code library. What should I do to let GPT help me understand the source code?

Can you support uploading GitHub repolinks?

14

Can you support uploading GitHub warehouse links? such as: https://github.com/panjh/dxflib.git

hope it could support azure openai key

1

as you know,Less restrictions on using GPT4 on Azure openai. so,hope code could support azure openai key future.

Is it possible to add ShuffleNetV2 as backbone in the official repo?

2

### Search before asking - [X] I have searched the YOLOv5 [issues](https://github.com/ultralytics/yolov5/issues) and [discussions](https://github.com/ultralytics/yolov5/discussions) and found no similar questions. ### Question According to the information I found, ShuffleNetV2 has achieved...