huapohen / general_parking_slot_detection Goto Github PK
View Code? Open in Web Editor NEWLicense: Apache License 2.0
License: Apache License 2.0
请问可以重新提供转换脚本吗
https://github.com/dohoseok/context-based-parking-slot-detect/
项目里没有转换脚本
大佬,Train.py跑不通呀,能否更新一版
您好, 想请教个问题, 如果车位的A点被截断(即位于图片之外), 这样的车位需要标注么? 另外算法里回归的是AB的长度, 所以个人理解, 如果B点位于图片之外, 这种车位也不标注对么?
也就是说, 本算法所要求的标注, 只有当A, B两点(无论是否遮挡)都位于图片内时才标注该车位, 对吗?
你好,请问ps2.0数据集怎么转化成ps2.0_convert呢。怎么转化成fullimg数据集呢
thanks for your wonderful work and open resource. Where to download the raw demo avm video without the slot detection result? thanks for you reply in advance.
PSV Resourse - TiEV research3 PSV
这个链接打不开,能不能提供一下云盘下载链接
您好,我想请教一下,为什么我用train.py进行训练,用的yolox_single_scale网络模型,保存下来的权重文件有180多MB呀,我看您这边提供的权重文件有的只有9MB大小,想知道您是怎么做到的
请问为什么要把图片灰度化后送入网络
前辈好,我通过全图训练的模型预测公开数据集发现效果很差,是不是因为用的推理代码是左右两部分的图导致的,可以提供一下全图训练对应的推理代码嘛
您好,我直接使用了您提供的fullimg权重预测ps2.0数据集(全图),预测结果不理想,重新训练了一组权重(在ps2.0上ap达到92%)预测的结果也没有改善,请问预测的时候有什么要调整的地方吗
大佬,请教下,我使用search_hyperparams.py生成的参数json文件,但载入后报错:
File "train.py", line 299, in
dataloaders = data_loader.fetch_dataloader(params)
File "/home/lto5sgh/2023-parking-slots/general_parking_slot_detection-master/dataset/data_loader.py", line 525, in fetch_dataloader
train_ds = AutoParkingData(params, mode="train")
File "/home/lto5sgh/2023-parking-slots/general_parking_slot_detection-master/dataset/data_loader.py", line 54, in init
percentage = int(len(imgs_file) * data_ratio[1])
ValueError: invalid literal for int() with base 10: 'eee'
我在Train.py中再次添加一遍
args.train_data_ratio = [["seoul", 1]]
args.val_data_ratio = [["seoul", 1]]
即可运行,但我添加的内容和json中是一致的,请问问题点在哪儿?如下是json的配置:
"train_data_ratio": [
"seoul",
1
],
"val_data_ratio": [
"seoul",
1
],
敬爱的前辈好,diy数据集的链接失效了,帮忙重新发一个链接哈
子图训练,斜方向车位分割线sin cos角度预测的不好,斜方向的车位类型预测出来基本都是与车位入口线垂直,但是ap精度确依然很高,ps2.0达到了99.54。
请问是训练集缺少这种图片的原因吗?
你好,非常感谢你的分享。想请教你一下,计算loss的时候,当deltanorm_or_direction是direction时,我看sepline的角度是用如下代码
x_ratio, y_ratio = 1, H / W
radian = math.atan2(marking_point.lenSepLine_y * y_ratio, marking_point.lenSepLine_x * x_ratio)
你能说明一下,为什么需要乘以y_ratio和x_ratio吗?
您好,请问一下,全图训练看了一下您提供的train.log. (自己训练也是) 随着训练的进行, loss不断降低.。为什么average_precision精度一直在跳变。
看到你的偏移是在归一化中的差值,可能就不是(-1,1)的均匀区间,反而可能是(-0.3,0.3)的区间。
而预测结果是用tanh函数激活的,这是不是结果不好的原因。
如果对偏移的长度单独归一化可以么?
或者说根本不是因为这个问题!而是你的图像缩嘚太小,网络模型结构不够复杂呢?
The previous link was blocked, and the new link is as follows:
https://pan.baidu.com/s/1wJKy1w1W6LLkQcG5fHQLDA?pwd=gpsd
A declarative, efficient, and flexible JavaScript library for building user interfaces.
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google ❤️ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.