davidfrz/yolov5_distance_count

使用yolov5，双目摄像头进行测距

yolov5-Binocular camera-distance count-ranging

2024年1月12日-- 更新一下进度 现在已经毕业啦 有空管理这个仓库了 所以在B站或者这里面提的新问题我看到的都会解答 我现在也有自己的时间了 在做一个AI集合的硬件内容 包括这些功能 也希望有合作可以看到我 Thanks！

中文文档 --->https://github.com/davidfrz/yolov5_distance_count/blob/master/README_CN.md

This project can get object recognition and distance display of the measured object through YOLOV5 target detection box with binocular camera.
Sample Vedio ===>[https://www.bilibili.com/video/BV1QK411w71d]

yolov5

All project is based on yolov5 ===>https://github.com/ultralytics/yolov5
And I put weight(yolov5s.pt) inside.

version

I chose the v3.1 version, but this project is just a matter of adding a few .py files, whatever version will do (at least for now).

HOW TO USE

I add three files in origin "yolov5" ---> camera_config.py dis_count.py video_remain.py

Files	instructions
camera_config.py	Binocular camera parameters
dis_count.py	Depth map, distance matrix
video_remain.py	main

RESULT

DEVICEE	FPS
1650	20
TX2	12
NX	15

JUST RUN video_remain.py and you can get what you want.

if this project can help you, give me a star,plzzzzzz!

Star History

yolov5_distance_count's Issues

已star，大佬我试了一下你的程序，目标检测没啥问题，但是深度图画面很花（我用matlab标定了参数），所以测距偏差比较大

⭐

AttributeError: 'Upsample' object has no attribute 'recompute_scale_factor'

单设备号的双目摄像头应该怎么接收参数呢

我看你的写法是cap1 cap2分别有参数接收的，单设备号的摄像头是不是不需要做分割直接使用呀

没有加载到数据集

Traceback (most recent call last):
File "video_remain.py", line 79, in
img = [letterbox(x, new_shape=640, auto=True)[0] for x in imgs]
File "video_remain.py", line 79, in
img = [letterbox(x, new_shape=640, auto=True)[0] for x in imgs]
File "/output/yolov5_distance_count/utils/datasets.py", line 735, in letterbox
shape = img.shape[:2] # current shape [height, width]
AttributeError: 'NoneType' object has no attribute 'shape'

非常感兴趣后续的分享，身体健康哦！

请问电脑只有cpu，出现这种情况咋弄

RuntimeError: Attempting to deserialize object on a CUDA device but torch.cuda.is_available() is False. If you are running on a CPU-only machine, please use torch.load with map_location=torch.device('cpu') to map your storages to the CPU.

如何以图片形式输入

请问up，要想输入左右两张图片来测距的话，应该咋搞啊

图像经过立体校正后图像有所差异

您好，我在实验过程中发现经过立体校正后的图像同当前摄像头所获取到的图像可能会存在差异，在边缘的物体可能会消失是这样的嘛

camera_configs.py 里面的平移矩阵是什么

如题 求怎么测量 是双目摄像头位置关系吗？

请教：我用yolov5 源码训练的权重在你的程序上报错，然后用你的代码训练 自定义数据集也各种报错，求指点

Traceback (most recent call last):
File "C:\Users\xl\AppData\Roaming\Python\Python38\site-packages\torch\serialization.py", line 594, in load
return _load(opened_zipfile, map_location, pickle_module, **pickle_load_args)
File "C:\Users\xl\AppData\Roaming\Python\Python38\site-packages\torch\serialization.py", line 853, in _load
result = unpickler.load()
AttributeError: Can't get attribute 'DetectionModel' on <module 'models.yolo' from 'D:\0000xl\pyProjects\distance\models\yolo.py'>

单设备号该咋改啊