diff --git a/docs/en/advanced_guides/codecs.md b/docs/en/advanced_guides/codecs.md
index 610bd83a57..7c98ba31d9 100644
--- a/docs/en/advanced_guides/codecs.md
+++ b/docs/en/advanced_guides/codecs.md
@@ -8,7 +8,7 @@ MMPose 1.0 introduced a new module **Codec** to integrate the encoding and decod
 
 Here is a diagram to show where the `Codec` is:
 
-![codec-en](https://user-images.githubusercontent.com/13503330/187112635-c01f13d1-a07e-420f-be50-3b8818524dec.png)
+![pose_estimator_en](https://github.com/open-mmlab/mmpose/assets/13503330/0764baab-41c7-4a1d-ab64-5d7f9dfc8eec)
 
 A typical codec consists of two parts:
 
diff --git a/docs/en/advanced_guides/implement_new_models.md b/docs/en/advanced_guides/implement_new_models.md
index 8d2809421e..da46a99e39 100644
--- a/docs/en/advanced_guides/implement_new_models.md
+++ b/docs/en/advanced_guides/implement_new_models.md
@@ -31,7 +31,7 @@ You can refer to the following flow chart to locate the module you need to imple
 
 In pose estimatiors, we will define the inference process of a model, and decode the model output results in `predict()`, first transform it from `output space` to `input image space` using the [codec](./codecs.md), and then combine the meta information to transform to `original image space`.
 
-![pose_estimator_en](https://github.com/open-mmlab/mmpose/assets/13503330/48c3813e-b977-4215-b5bc-e7379cfd2bce)
+![pose_estimator_en](https://github.com/open-mmlab/mmpose/assets/13503330/0764baab-41c7-4a1d-ab64-5d7f9dfc8eec)
 
 Currently, MMPose supports the following types of pose estimator:
 
diff --git a/docs/en/guide_to_framework.md b/docs/en/guide_to_framework.md
index bb1efed08f..fceb37a142 100644
--- a/docs/en/guide_to_framework.md
+++ b/docs/en/guide_to_framework.md
@@ -298,7 +298,7 @@ In a keypoint detection task, data will be transformed among three scale spaces:
 
 Here is a diagram to show the workflow of data transformation among the three scale spaces:
 
-![migration-en](https://user-images.githubusercontent.com/13503330/187190213-cad87b5f-0a95-4f1f-b722-15896914ded4.png)
+![tour_en](https://github.com/open-mmlab/mmpose/assets/13503330/e82710e6-4181-4eb0-8185-7075b43dbec3)
 
 In MMPose, the modules used for data transformation are under `[$MMPOSE/mmpose/datasets/transforms](https://github.com/open-mmlab/mmpose/tree/main/mmpose/datasets/transforms)`, and their workflow is shown as follows:
 
diff --git a/docs/zh_cn/advanced_guides/codecs.md b/docs/zh_cn/advanced_guides/codecs.md
index 85d4d2e54b..60c588a239 100644
--- a/docs/zh_cn/advanced_guides/codecs.md
+++ b/docs/zh_cn/advanced_guides/codecs.md
@@ -8,7 +8,7 @@ MMPose 1.0 中引入了新模块 **编解码器（Codec）** ，将关键点数
 
 编解码器在工作流程中所处的位置如下所示：
 
-![codec-cn](https://user-images.githubusercontent.com/13503330/187829784-4d5939de-97d7-43cc-b934-c6d17c02d589.png)
+![pose_estimator_cn](https://github.com/open-mmlab/mmpose/assets/13503330/0c048f66-b889-4268-937f-71b8753b505f)
 
 一个编解码器主要包含两个部分：
 
diff --git a/docs/zh_cn/advanced_guides/implement_new_models.md b/docs/zh_cn/advanced_guides/implement_new_models.md
index e233850277..d3ed96bd37 100644
--- a/docs/zh_cn/advanced_guides/implement_new_models.md
+++ b/docs/zh_cn/advanced_guides/implement_new_models.md
@@ -30,7 +30,7 @@ mmpose
 
 在姿态估计范式中，我们会定义一个模型的推理流程，并在 `predict()` 中对模型输出结果进行解码，先将其从 `输出尺度空间` 用 [编解码器](./codecs.md) 变换到 `输入图片空间`，然后再结合元信息变换到 `原始图片空间`。
 
-![image](https://github.com/open-mmlab/mmpose/assets/13503330/e3e700ac-a047-4cff-9017-67f83676b8cb)
+![pose_estimator_cn](https://github.com/open-mmlab/mmpose/assets/13503330/0c048f66-b889-4268-937f-71b8753b505f)
 
 当前 MMPose 已支持以下几类算法范式：
 
diff --git a/docs/zh_cn/guide_to_framework.md b/docs/zh_cn/guide_to_framework.md
index 132233fd5d..b4c44192a5 100644
--- a/docs/zh_cn/guide_to_framework.md
+++ b/docs/zh_cn/guide_to_framework.md
@@ -296,7 +296,7 @@ test_pipeline = [
 
 数据在三个空间中变换的流程如图所示：
 
-![migration-cn](https://user-images.githubusercontent.com/13503330/187831574-13804daf-f498-47c2-ba43-64b8e6ffe3dd.png)
+![tour_cn](https://github.com/open-mmlab/mmpose/assets/13503330/4c989d86-e824-49ea-9ba8-b3978548db37)
 
 在MMPose中，数据变换所需要的模块在 `[$MMPOSE/mmpose/datasets/transforms](https://github.com/open-mmlab/mmpose/tree/main/mmpose/datasets/transforms)` 目录下，它们的工作流程如图所示：