diff --git a/perception/traffic_light_classifier/README.md b/perception/traffic_light_classifier/README.md
index 24928c9ee4de0..b598d877d5293 100644
--- a/perception/traffic_light_classifier/README.md
+++ b/perception/traffic_light_classifier/README.md
@@ -87,6 +87,70 @@ These colors and shapes are assigned to the message as follows:
 | `red_max_s`    | int  | the maximum saturation of red color            |
 | `red_max_v`    | int  | the maximum value (brightness) of red color    |
 
+## Customization of CNN model
+
+Currently, in Autoware, [MobileNetV2](https://arxiv.org/abs/1801.04381v3) is used as CNN classifier by default. The corresponding onnx file is [data/traffic_light_classifier_mobilenetv2.onnx](./data/traffic_light_classifier_mobilenetv2.onnx).
+Also, you can apply the following models shown as below, for example.
+
+- [EfficientNet](https://arxiv.org/abs/1905.11946v5)
+- [ResNet](https://openaccess.thecvf.com/content_cvpr_2016/html/He_Deep_Residual_Learning_CVPR_2016_paper.html)
+- [MobileNetV3](https://arxiv.org/abs/1905.02244)
+  ...
+
+In order to train models and export onnx model, we recommend [open-mmlab/mmclassification](https://github.com/open-mmlab/mmclassification.git).
+Please follow the [official document](https://mmclassification.readthedocs.io/en/latest/) to install and experiment with mmclassification. If you get into troubles, [FAQ page](https://mmclassification.readthedocs.io/en/latest/faq.html) would help you.
+
+The following steps are example of a quick-start.
+
+### step 0. Install [MMCV](https://github.com/open-mmlab/mmcv.git) and [MIM](https://github.com/open-mmlab/mim.git)
+
+_NOTE_ : First of all, install [PyTorch](https://pytorch.org/) suitable for your CUDA version (CUDA11.6 is supported in Autoware).
+
+In order to install mmcv suitable for your CUDA version, install it specifying a url.
+
+```shell
+# Install mim
+$ pip install -U openmim
+
+# Install mmcv on a machine with CUDA11.6 and PyTorch1.13.0
+$ pip install mmcv-full -f https://download.openmmlab.com/mmcv/dist/cu116/torch1.13/index.html
+```
+
+### step 1. Install MMClassification
+
+You can install MMClassification as a Python package or from source.
+
+```shell
+# As a Python package
+$ pip install mmcls
+
+# From source
+$ git clone https://github.com/open-mmlab/mmclassification.git
+$ cd mmclassification
+$ pip install -v -e .
+```
+
+### step 2. Train your model
+
+Train model with your experiment configuration file. For the details of config file, see [here](https://mmclassification.readthedocs.io/en/latest/tutorials/config.html).
+
+```shell
+# [] is optional, you can start training from pre-trained checkpoint
+$ mim train mmcls YOUR_CONFIG.py [--resume-from YOUR_CHECKPOINT.pth]
+```
+
+### step 3. Export onnx model
+
+In exporting onnx, use `mmclassificatoin/tools/deployment/pytorch2onnx.py` or [open-mmlab/mmdeploy](https://github.com/open-mmlab/mmdeploy.git).
+
+```shell
+cd ~/mmclassification/tools/deployment
+python3 pytorch2onnx.py YOUR_CONFIG.py ...
+```
+
+After obtaining your onnx model, update parameters defined in the launch file (e.g. `model_file_path`, `label_file_path`, `input_h`, `input_w`...).
+Note that, we only support labels defined in [autoware_auto_perception_msgs::msg::TrafficLight](https://github.com/tier4/autoware_auto_msgs/blob/tier4/main/autoware_auto_perception_msgs/msg/TrafficLight.idl).
+
 ## Assumptions / Known limits
 
 <!-- Write assumptions and limitations of your implementation.