GithubHelp home page GithubHelp logo

patta's Introduction

patta's People

Contributors

ainavo avatar gt-zhangacer avatar siguremo avatar whitefirefox avatar wongziseoi avatar

Stargazers

 avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar

Watchers

 avatar  avatar  avatar  avatar  avatar

patta's Issues

Binary string

显示无法从txt读取图片路径是为什么?对txt格式有什么要求吗?
ValueError: (InvalidArgument) Failed to parse program_desc from binary string.
[Hint: Expected desc_.ParseFromString(binary_str) == true, but received desc_.ParseFromString(binary_str):0 != true:1.] (at /paddle/paddle/fluid/framework/program_desc.cc:143)

【PaddlePaddle Hackathon】98 搜索测试图像增强最佳方案探索

(此 ISSUE 为 PaddlePaddle Hackathon 活动的任务 ISSUE,更多详见PaddlePaddle Hackathon

PaTTA 就是一个致力于让模型表现更加稳定的飞桨模型测试增强工具箱,其原理为在测试时对要推理的数据进行增强,通过投票形式选出更稳健的推理结果。

【任务说明】

  • 任务标题:搜索测试图像增强最佳方案探索

  • 技术标签:Python、PaddlePaddle

  • 任务难度:简单

  • 详细描述:在一般的深度学习赛事中,模型融合、TTA 等策略虽然能有效提升选手成绩,但这些方案在性能上往往难以应用于真实场景。虽然 PaTTA 提供了 TTA 工具,但我们也可以思考是否可以通过统计等方式,在用户预测单张图像时尽可能推荐出一个推理性能均衡点,在较低的速度影响下依旧可以提升模型效果。在这个项目中,需要你在同样环境下,在 Cifar100 数据集上进行推理,做到速度影响在 5% 以内,精度仍可具备至少 0.1% 的提升。

【提交内容】

  • 项目 PR 到 PaTTA

  • 技术说明文档

【技术要求】

  • 可跑通 PaddlePaddle 核心框架下任一图像分类任务

【PaddlePaddle Hackathon】96 图像分类模型解释性可视化探究

(此 ISSUE 为 PaddlePaddle Hackathon 活动的任务 ISSUE,更多详见PaddlePaddle Hackathon

PaTTA 是一个致力于让模型表现更加稳定的飞桨模型测试增强工具箱。

【任务说明】

  • 任务标题:图像分类模型解释性可视化探究

  • 技术标签:PaTTA、Python、PaddlePaddle

  • 任务难度:简单

详细描述:深度学习模型在结构上很难具备“可解释”能力,然而这并不影响我们通过梯度、噪音等方式去解释模型到底在关注什么,也就意味着我们在一些比赛中也可以从通过该方式来了解模型的“关注点”从而提升比赛成绩。

在这个任务中,你需要从产品设计出发,也可以考虑如何优化可解释型算法,目的是将解释性工具箱 InterpretDL 或者自己实现的可解释性模块加入 PaTTA 工具箱中,为模型分析提供更多可能,使得用户在使用 PaTTA 工具箱进行推理结果增强时,可以通过简单的方式调用可视化解释性功能,向使用者提供解释性分析情况。

PaTTA 主页:https://github.com/AgentMaker/PaTTA

InterpretDL 主页:https://github.com/PaddlePaddle/InterpretDL

【提交内容】

  • 项目 PR 到 PaTTA
  • 技术说明文档

【技术要求】

  • 具有基础的 Python 开发能力

  • 有使用 Matplotlib 或 OpenCV 等任一 Python 图像库的使用经历

【PaddlePaddle Hackathon】97 新增图像数据增强算法

(此 ISSUE 为 PaddlePaddle Hackathon 活动的任务 ISSUE,更多详见PaddlePaddle Hackathon

PaTTA 是一个致力于让模型表现更加稳定的飞桨模型测试增强工具箱,其原理为在测试时对要推理的数据进行增强,通过投票形式选出更稳健的推理结果。

【任务说明】

  • 任务标题:新增图像数据增强算法

  • 技术标签:Python

  • 任务难度:简单

  • 详细描述:数据增强是一种比较有效的模型能力提升方式,更多的组合可使得模型在训练时更加关注目标特征,从而进一步提升模型成绩。目前 PaTTA 中仅具备高频的图像数据增强算法。本这个项目,需要你新增不低于5个图像方向的数据增强算法,并且这些算法能够略微、显著提升推理成绩,以提升 PaTTA 可用性。

【提交内容】

  • 项目 PR 到 PaTTA

  • 技术说明文档

【项目技术要求】

  • 具有基础的 Python 开发能力

  • 有过在深度学习中使用图像增强的经历

preprocess issue

issue 1

当我将crop_size调至(1024,512), 报错

Traceback (most recent call last):
File "PaTTA/tools/seg.py", line 41, in
main(args.batch_size, imgs_list, args.crop_size)
File "PaTTA/tools/seg.py", line 26, in main
tensor_img = tta_model(tensor_img)
File "/opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages/paddle/fluid/dygraph/layers.py", line 902, in call
outputs = self.forward(*inputs, **kwargs)
File "/opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages/patta/wrappers.py", line 39, in forward
augmented_output = self.model(augmented_image, *args)[0]
File "/opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages/paddle/fluid/dygraph/layers.py", line 902, in call
outputs = self.forward(*inputs, **kwargs)
File "/opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages/paddle/fluid/dygraph/io.py", line 1170, in i_m_p_l
return _run_dygraph(self, input, program_holder)
File "/opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages/paddle/fluid/dygraph/io.py", line 733, in _run_dygraph
'is_test': instance._is_test
File "/opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages/paddle/fluid/dygraph/tracer.py", line 45, in trace_op
not stop_gradient)
ValueError: (InvalidArgument) Broadcast dimension mismatch. Operands could not be broadcast together with the shape of X = [16, 48, 128, 256] and the shape of Y = [16, 48, 384, 384]. Received [128] in X is not equal to [384] in Y at i:2.
[Hint: Expected x_dims_array[i] == y_dims_array[i] || x_dims_array[i] <= 1 || y_dims_array[i] <= 1 == true, but received x_dims_array[i] == y_dims_array[i] || x_dims_array[i] <= 1 || y_dims_array[i] <= 1:0 != true:1.] (at /paddle/paddle/fluid/operators/elementwise/elementwise_op_function.h:160)
[operator < elementwise_add > error] [operator < run_program > error]

事实上修改任意crop_size都报错,但是改为1536,1536即数据集的图片尺寸,上述错误解决,但是issue2出现

issue 2

Traceback (most recent call last):
File "PaTTA/tools/seg.py", line 41, in
main(args.batch_size, imgs_list, args.crop_size)
File "PaTTA/tools/seg.py", line 26, in main
tensor_img = tta_model(tensor_img)
File "/opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages/paddle/fluid/dygraph/layers.py", line 902, in call
outputs = self.forward(*inputs, **kwargs)
File "/opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages/patta/wrappers.py", line 39, in forward
augmented_output = self.model(augmented_image, *args)[0]
File "/opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages/paddle/fluid/dygraph/layers.py", line 902, in call
outputs = self.forward(*inputs, **kwargs)
File "/opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages/paddle/fluid/dygraph/io.py", line 1170, in i_m_p_l
return _run_dygraph(self, input, program_holder)
File "/opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages/paddle/fluid/dygraph/io.py", line 733, in _run_dygraph
'is_test': instance._is_test
File "/opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages/paddle/fluid/dygraph/tracer.py", line 45, in trace_op
not stop_gradient)
ValueError: (InvalidArgument) The 'shape' in ReshapeOp is invalid. The input tensor X'size must be equal to the capacity of 'shape'. But received X's shape = [16, 512, 384, 384], X's size = 1207959552, 'shape' is [1, 512, 147456], the capacity of 'shape' is 75497472.
[Hint: Expected capacity == in_size, but received capacity:75497472 != in_size:1207959552.] (at /paddle/paddle/fluid/operators/reshape_op.cc:222)
[operator < reshape2 > error] [operator < run_program > error]

【PaddlePaddle Hackathon】AgentMaker 任务合集

Hi,大家好,非常高兴的告诉大家,首届 PaddlePaddle Hackathon 开始啦。PaddlePaddle Hackathon 是面向全球开发者的深度学习领域编程活动,鼓励开发者了解与参与 PaddlePaddle。本次共有四大方向(PaddlePaddle、Paddle Family、Paddle Friends、Paddle Anything)四大方向,共计100个任务共大家完成。详细信息可以参考 PaddlePaddle Hackathon 说明。大家是否已经迫不及待了呢~

本 ISSUE 是 Paddle Friends 专区 AgentMaker 方向任务合集。具体任务列表如下:

序号 难度 任务 ISSUE
96 ⭐️ 【PaddlePaddle Hackathon】96 图像分类模型解释性可视化探究
97 ⭐️ 【PaddlePaddle Hackathon】97 新增图像数据增强算法
98 ⭐️ 【PaddlePaddle Hackathon】98 搜索测试图像增强最佳方案探索
99 ⭐️ 【PaddlePaddle Hackathon】99 为 AgentOCR 工具适配 JavaScript 环境
100 ⭐️ ⭐️ 【PaddlePaddle Hackathon】100 制作 Rubick 深度学习相关小插件

若想要认领本次活动任务,请至 PaddlePaddle Hackathon Pinned ISSUE 完成活动报名以及任务认领。

活动官网:PaddlePaddle Hackathon

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. 📊📈🎉

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google ❤️ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.