Encoding images are slightly different than encoding format for other files

未关闭
#1,663 0 条评论 0 个 reaction 已指派 0 人 在 GitHub 查看

还没有人认领这个 Issue。

评估

难度
4/5
预计耗时
3-5 天
新手友好度
30/100
Issue 类型
缺陷
描述清晰度
需要澄清
活跃度
停滞
技术栈
python
领域
computer-vision

调研方向

Start with the linked Colab reproduction and compare the bytes produced by the image encoding operation with those written through tf.io.write_file. Check the REST API path described in the issue and establish whether the returned bytes can be written as png, jpeg, bmp, or gif files. Done means the discrepancy is explained and a reproducible correction or documented limitation is identified.

由索引模型根据 Issue 内容生成。

描述

I am trying to write a tensor to a png, jpeg, bmp, or gif. As seen in colab in bottom, when I take the encoded bytes and write it a file using python, it doesn't work. Trying opening the image it gets corrupted
When I use tf.io.write_file it works. I also tried taking a png and using tf.io_decode_png and than tf.io.encoding_png and the image string bytes differ from the original png. I am using tensorflow serving and using this png result through the rest api, where I don't have access to tf.io.write_file so I was wondering what encode_png does differently, than traditional png encodings.

https://www.loom.com/share/4b32b330a0e742cab2ea3a1a9af11b22
https://colab.research.google.com/drive/13Y4ET8jAHZDq7qwUYgg8nuBkxqniUqGu#scrollTo=IcXVZepnhkW4

主要语言
C++
星标
739
派生
311
30 天内没有已合并 PR

贡献指南

打开贡献指南

从这里开始

  1. 先读完整个 Issue，再读项目的贡献指南。
  2. 在 Issue 下留言说明你要接手 —— 这能避免两个人做同样的事。
  3. Fork 仓库，在一个分支上完成修改。
  4. 提交 Pull Request，并在描述里引用这个 Issue 编号。

tensorflow/io 的其他 Issue

查看 tensorflow/io 的全部 Issue

相似的 Issue

更多 C++ Issue

把新 issue 发到你的邮箱

精选适合新手参与的 GitHub issue 摘要。