lenonhere

How to train your own object detection models using the TensorFlow Object Detection API (2020 Update)

This started as a summary of this nice tutorial, but has since then become its own thing.

Modify index.txt with your output path and proxy info
Use Text Builder to build configuration for Surge: $ text-builder -index /path/to/index.txt Or run $ sh build-all to build all your index files.
Import configuration via AirDrop/iTunes/Dropbox/iCloud

本人不提供任何保证和技术支持，使用者自负风险。
There are no guarantees, no any support. Use it at your own risk.

基于我自身浏览习惯＋深圳电信线路的namebench测试，整合了以下公共DNS列表的测试结果：

个人结论是：无论使用国内还是国外的DNS，都在解析网站时受到相当的干扰，如果不是劫持和黑洞，就是纯粹的丢包超时。只有在用户端设好工具，所有海外DNS请求走加密通道，才可以杜绝触发黑洞和过滤。在用户端想办法过滤虚假的DNS结果也可行，但并不能真正解决超时的使用问题。

	SELECT table,
	formatReadableSize(sum(bytes)) as size,
	min(min_date) as min_date,
	max(max_date) as max_date
	FROM system.parts
	WHERE active
	GROUP BY table

	'''This script goes along the blog post
	"Building powerful image classification models using very little data"
	from blog.keras.io.
	It uses data that can be downloaded at:
	https://www.kaggle.com/c/dogs-vs-cats/data
	In our setup, we:
	- created a data/ folder
	- created train/ and validation/ subfolders inside data/
	- created cats/ and dogs/ subfolders inside train/ and validation/
	- put the cat pictures index 0-999 in data/train/cats

	'''This script goes along the blog post
	"Building powerful image classification models using very little data"
	from blog.keras.io.
	It uses data that can be downloaded at:
	https://www.kaggle.com/c/dogs-vs-cats/data
	In our setup, we:
	- created a data/ folder
	- created train/ and validation/ subfolders inside data/
	- created cats/ and dogs/ subfolders inside train/ and validation/
	- put the cat pictures index 0-999 in data/train/cats

	[General]
	#我的注释都是 # 开头，所以如果你用 vim，直接 :g/^#/d 就可以一次性清除所有注释
	#开头这段skip包含以下几个目的：1，私网IP跳过，提高内网性能；2，苹果的一些服务跳过，比如公共热点wifi要先测试captive.apple.com。可解决很多内网的 TCP毛病，例如 kodi remote软件无法遥控
	skip-proxy = 10.0.0.0/8,169.254/16,172.16.0.0/12,192.168.0.0/16,224.0.0.0/4, localhost, .local,api.smoot.apple.com,configuration.apple.com,xp.apple.com,smp-device-content.apple.com,guzzoni.apple.com,captive.apple.com,.ess.apple.com,.push.apple.com,.push-apple.com.akadns.net
	#下面这一段则是完全跳过 Surge，最重要的一个是让 UDP包可以传输，解决很多内网毛病，例如DLNA，NFS或者btsync等组播类/UDP类应用
	bypass-tun = 10.0.0.0/8, 169.254.0.0/16, 172.16.0.0/12, 192.168.0.0/16, 224.0.0.0/4, 0.0.0.0/8, 1.0.0.0/9, 1.160.0.0/11, 1.192.0.0/11, 10.0.0.0/8, 14.0.0.0/11, 14.96.0.0/11, 14.128.0.0/11, 14.192.0.0/11, 27.0.0.0/10, 27.96.0.0/11, 27.128.0.0/9, 36.0.0.0/10, 36.96.0.0/11, 36.128.0.0/9, 39.0.0.0/11, 39.64.0.0/10, 39.128.0.0/10, 42.0.0.0/8, 43.224.0.0/11, 45.64.0.0/10, 47.64.0.0/10, 49.0.0.0/9, 49.128.0.0/11, 49.192.0.0/10, 54.192.0.0/11, 58.0.0.0/9, 58.128.0.0/11, 58.192.0.0/10, 59.32.0.0/11, 5

	import (
	"crypto/md5"
	"encoding/hex"
	)

	func GetMD5Hash(text string) string {
	hasher := md5.New()
	hasher.Write([]byte(text))
	return hex.EncodeToString(hasher.Sum(nil))
	}