# 图片转为文字描述（自然语言）

ComfyUI-ImageToText.json

@@ -0,0 +1,152 @@
+{
+  "last_node_id": 4,
+  "last_link_id": 2,
+  "nodes": [
+    {
+      "id": 3,
+      "type": "ComfyUI_ImageToText",
+      "pos": [
+        726,
+        120
+      ],
+      "size": {
+        "0": 315,
+        "1": 58
+      },
+      "flags": {},
+      "order": 1,
+      "mode": 0,
+      "inputs": [
+        {
+          "name": "images",
+          "type": "IMAGE",
+          "link": 1,
+          "label": "images"
+        }
+      ],
+      "outputs": [
+        {
+          "name": "text_positive",
+          "type": "STRING",
+          "links": [
+            2
+          ],
+          "shape": 3,
+          "label": "text_positive",
+          "slot_index": 0
+        }
+      ],
+      "properties": {
+        "Node name for S&R": "ComfyUI_ImageToText"
+      },
+      "widgets_values": [
+        "Yes"
+      ]
+    },
+    {
+      "id": 2,
+      "type": "LoadImage",
+      "pos": [
+        282,
+        113
+      ],
+      "size": {
+        "0": 315,
+        "1": 314
+      },
+      "flags": {},
+      "order": 0,
+      "mode": 0,
+      "outputs": [
+        {
+          "name": "IMAGE",
+          "type": "IMAGE",
+          "links": [
+            1
+          ],
+          "shape": 3,
+          "label": "图像",
+          "slot_index": 0
+        },
+        {
+          "name": "MASK",
+          "type": "MASK",
+          "links": null,
+          "shape": 3,
+          "label": "遮罩"
+        }
+      ],
+      "properties": {
+        "Node name for S&R": "LoadImage"
+      },
+      "widgets_values": [
+        "00489-MexxL_LCM2_YY-1034089384-1-960-20240111205721.jpg",
+        "image"
+      ]
+    },
+    {
+      "id": 4,
+      "type": "ShowText|pysssss",
+      "pos": [
+        1133.3125,
+        133.7734375
+      ],
+      "size": [
+        380.328125,
+        239.95703125
+      ],
+      "flags": {},
+      "order": 2,
+      "mode": 0,
+      "inputs": [
+        {
+          "name": "text",
+          "type": "STRING",
+          "link": 2,
+          "widget": {
+            "name": "text"
+          },
+          "label": "文本"
+        }
+      ],
+      "outputs": [
+        {
+          "name": "STRING",
+          "type": "STRING",
+          "links": null,
+          "shape": 6,
+          "label": "字符串"
+        }
+      ],
+      "properties": {
+        "Node name for S&R": "ShowText|pysssss"
+      },
+      "widgets_values": [
+        "",
+        "The image depicts a young girl with black hair seated at a desk, intently focused on her laptop. The desk is adorned with a rainbow-colored record player and a rainbow-colored music note, adding a vibrant touch to the scene. The background is a stark white, providing a contrast that makes the girl and her surroundings stand out. The image is framed by a black border, and a watermark in the bottom right corner reads \"© 2020\"."
+      ]
+    }
+  ],
+  "links": [
+    [
+      1,
+      2,
+      0,
+      3,
+      0,
+      "IMAGE"
+    ],
+    [
+      2,
+      3,
+      0,
+      4,
+      0,
+      "STRING"
+    ]
+  ],
+  "groups": [],
+  "config": {},
+  "extra": {},
+  "version": 0.4
+}

ImageToText.py

@@ -0,0 +1,51 @@
+from transformers import AutoModelForCausalLM, AutoTokenizer
+from PIL import Image
+
+import numpy as np
+
+model_id = "vikhyatk/moondream2"
+revision = "2024-04-02"
+
+class ComfyUI_ImageToText:
+    def __init__(self):
+        pass
+
+    @classmethod
+    def INPUT_TYPES(s):
+        return {
+            "required": {
+                "images": ("IMAGE",),
+                "log_prompt": (["No", "Yes"], {"default":"Yes"}),
+            },
+        }
+
+    RETURN_TYPES = ('STRING',)
+    RETURN_NAMES = ('text_positive',)
+    FUNCTION = "image2text"
+    OUTPUT_NODE = True
+    CATEGORY = "ComfyUI_Mexx"
+
+    def image2text(self, images, log_prompt):
+        pil_images = []
+        for image in images:
+            i = 255. * image.cpu().numpy()
+            img = Image.fromarray(np.clip(i, 0, 255).astype(np.uint8))
+            pil_images.append(img)
+        image = pil_images[0]
+        model = AutoModelForCausalLM.from_pretrained(
+            model_id, trust_remote_code=True, revision=revision
+        )
+        tokenizer = AutoTokenizer.from_pretrained(model_id, revision=revision)
+        enc_image = model.encode_image(image)
+        en = model.answer_question(enc_image, "Describe this image.", tokenizer)
+        if log_prompt == "Yes":
+            print(f"ImageToText: {en}")
+        return [en]
+
+NODE_CLASS_MAPPINGS = {
+    "ComfyUI_ImageToText": ComfyUI_ImageToText
+}
+
+NODE_DISPLAY_NAME_MAPPINGS = {
+    "ComfyUI_ImageToText": "ComfyUI_ImageToText"
+}

README.md

@@ -0,0 +1,18 @@
+# ComfyUI_ImageToText
+
+## 功能简述
+
+把图片以自然语言描述出来.
+
+## 使用图例
+
+![demo.png](image%2Fdemo.png)
+
+## 工作流举例
+
+SDXL模型下载地址(欢迎点赞点关注): https://www.liblib.art/modelinfo/5913fb0765ce4a4ba210cb1c898df276
+工作流文件（直接拖拽到ComfyUI的页面里即可）: [ComfyUI-ImageToText.json](ComfyUI-ImageToText.json)
+
+## 使用了的模型
+
+https://huggingface.co/vikhyatk/moondream2

__init__.py

@@ -0,0 +1,3 @@
+from .ImageToText import NODE_CLASS_MAPPINGS, NODE_DISPLAY_NAME_MAPPINGS
+
+__all__ = ['NODE_CLASS_MAPPINGS', 'NODE_DISPLAY_NAME_MAPPINGS']

requirements.txt

@@ -0,0 +1,3 @@
+transformers
+timm
+einops