Update app.py

app.py

@@ -7,10 +7,19 @@ from dotenv import load_dotenv
 import google.generativeai as genai
 import os
 
-load_dotenv()
-GOOGLE_API_KEY = os.getenv("GOOGLE_API")
-genai.configure(api_key=GOOGLE_API_KEY)
-model_vision = genai.GenerativeModel('gemini-pro-vision')
+from openai import OpenAI
+client = OpenAI(api_key="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")
+import base64
+
+# Open the image file and encode it as a base64 string
+def encode_image(image_path):
+    with open(image_path, "rb") as image_file:
+        return base64.b64encode(image_file.read()).decode("utf-8")
+
+# load_dotenv()
+# GOOGLE_API_KEY = os.getenv("GOOGLE_API")
+# genai.configure(api_key=GOOGLE_API_KEY)
+# model_vision = genai.GenerativeModel('gemini-pro-vision')
 
 def gemini_response_vision(input_texts, image):
     try:
@@ -59,8 +68,26 @@ def shot(image, labels_text, model_name, hypothesis_template_prefix, hypothesis_
         domains = [domain.strip(" ") for domain in domains_text.strip(" ").split(",")]
     else:
         #img = Image.open(image)
-        input_text = "Please describe the image from six dimensions, including weather (clear, sandstorm, foggy, rainy, snowy), angle (front, left, top), time (daytime, night), occlusion (unoccluded, lightly-occluded, partially-occluded, moderately-occluded, heavily-occluded), season (spring-summer, autumn, winter). Each dimension should be described in no more than 4 words and should match the image content. Please try to output from the options in the previous brackets. If there is no suitable result, output N/A."# Please also output a probability of your inference."# If there is no information in a certain dimension, you can directly output no information.
-        domains = gemini_response_vision(input_texts=input_text, image=image)
+        #input_text = "Please describe the image from six dimensions, including weather (clear, sandstorm, foggy, rainy, snowy), angle (front, left, top), time (daytime, night), occlusion (unoccluded, lightly-occluded, partially-occluded, moderately-occluded, heavily-occluded), season (spring-summer, autumn, winter). Each dimension should be described in no more than 4 words and should match the image content. Please try to output from the options in the previous brackets. If there is no suitable result, output N/A."# Please also output a probability of your inference."# If there is no information in a certain dimension, you can directly output no information.
+        #domains = gemini_response_vision(input_texts=input_text, image=image)
+        #IMAGE_PATH = './reasoning_xy.jpg'
+        base64_image = encode_image(image)
+        prompt = "Please describe the image from six dimensions, including weather (clear, sandstorm, foggy, rainy, snowy), angle (front, left, top), time (daytime, night), occlusion (unoccluded, lightly-occluded, partially-occluded, moderately-occluded, heavily-occluded), season (spring-summer, autumn, winter). Each dimension should be described in no more than 4 words and should match the image content. Please try to output from the options in the previous brackets. If there is no suitable result, output N/A."# Please also output a probability of your inference."# If there is no information in a certain dimension, you can directly output no information."
+        
+        response = client.chat.completions.create(
+            model="o1-preview",
+            messages=[
+                # {"role": "system", "content": "You are a helpful assistant that responds in Markdown. Help me with my math homework!"},
+                {"role": "user", "content": [
+                    {"type": "text", "text": prompt},
+                    {"type": "image_url", "image_url": {
+                        "url": f"data:image/png;base64,{base64_image}"}
+                    }
+                ]}
+            ],
+            temperature=0.0,
+        )
+        domains = response.choices[0].message.content
         print(domains)
         
     hypothesis_template = hypothesis_template_prefix + ' ' + hypothesis_template_suffix.format(*domains)