From 156c4eb4f2f6d5ac917f53b7091841dfff66a077 Mon Sep 17 00:00:00 2001
From: Koki Shinjo <sktometometo@gmail.com>
Date: Tue, 9 Jan 2024 23:47:39 +0900
Subject: [PATCH] [gpt4v_vqa] update README.md

---
 gpt4v_vqa/README.md | 10 +++++++++-
 1 file changed, 9 insertions(+), 1 deletion(-)

diff --git a/gpt4v_vqa/README.md b/gpt4v_vqa/README.md
index 4e6586cda1..a5573726da 100644
--- a/gpt4v_vqa/README.md
+++ b/gpt4v_vqa/README.md
@@ -11,13 +11,21 @@ catkin build gpt4v_vqa
 ## Usage
 
 ```bash
-roslaunch gpt4v_vqa vqa.launch api_key:=<YOUR_API_KEY>
+roslaunch gpt4v_vqa vqa.launch api_key:=<YOUR_API_KEY> VQA_INPUT_IMAGE:=<IMAGE TOPIC>
+```
+
+And from other terminal
+
+```bash
+$ rosrun gpt4v_vqa vqa_interpreter.py
 ```
 
 ## Nodes
 
 ### gpt4v_vqa
 
+This node is a ROS wrapper for GPT4V model. Its behavior is similar to [VQA node](../jsk_perception/node_scripts/vqa_node.py). But there is a difference that this node does not support continuous inference. This node use API only when action server is called.
+
 #### Subscribed Topics
 
 * **`~image`** ([sensor_msgs/Image])