Create app.py

app.py

@@ -0,0 +1,159 @@
+import gradio as gr
+import base64
+import requests
+import random
+import os
+from openai import OpenAI
+from PIL import Image
+import json
+import cohere
+
+iucn_api_token = os.environ.get('IUCN_API')
+cohere_api_token = os.environ.get('COHERE_API')
+openai_api_token = os.environ.get('OPENAI_API')
+
+client = OpenAI(api_key=openai_api_token)
+co = cohere.Client(cohere_api_token)
+
+def encode_image(image_path):
+  with open(image_path, "rb") as image_file:
+    return base64.b64encode(image_file.read()).decode('utf-8')
+
+def summarize_with_llm(text, prompt, max_token=210):
+  response = co.generate(
+    model='command',
+    prompt=f'This is a piece of information about an animal: "{text}". {prompt}',
+    max_tokens=max_token,
+    temperature=0.5,
+    k=0,
+    stop_sequences=[],
+    return_likelihoods='NONE')
+  return response.generations[0].text
+
+def get_iucn_data(genus, species):
+    iucn_narrative = requests.get(f"https://apiv3.iucnredlist.org/api/v3/species/narrative/{genus}%20{species}?token={iucn_api_token}")
+    iucn_status = requests.get(f"https://apiv3.iucnredlist.org/api/v3/species/history/name/{genus}%20{species}?token={iucn_api_token}")
+    iucn_common_name = requests.get(f"https://apiv3.iucnredlist.org/api/v3/species/common_names/{genus}%20{species}?token={iucn_api_token}")
+    iucn_web_link = requests.get(f"https://apiv3.iucnredlist.org/api/v3/weblink/{genus}%20{species}")
+
+    if iucn_narrative.status_code == 200:
+        narratives = iucn_narrative.json()
+        conservation_status = iucn_status.json()
+        if conservation_status['result'] == []:
+            return dict()
+            
+        status_category = conservation_status['result'][0]['category']
+        status_code = conservation_status['result'][0]['code']
+        common_name = iucn_common_name.json()['result'][0]['taxonname']
+        web_link = iucn_web_link.json()['rlurl']
+        threats = summarize_with_llm(narratives['result'][0]['threats'], 'In one sentence, the threats posing this species are', max_token=210)
+        population = summarize_with_llm(narratives['result'][0]['population'], 'In one sentence, estimation of the population of this species is', max_token=210)
+        habitat = summarize_with_llm(narratives['result'][0]['habitat'], 'Description of the habitat of this species is')
+        return {
+            "status_category": status_category,
+            "status_code": status_code,
+            "common_name": common_name,
+            "web_link": web_link,
+            "threats": threats.strip().split('.')[0],
+            "population": population.strip().split('.')[0],
+            "habitat": habitat.strip().split('.')[0]
+        }
+    else:
+        return dict()
+
+def get_taxonomy(image):
+    # Path to your image
+    id = random.randint(0, 1000)
+    image_path = f"upload_{id}.png"
+    image.save(image_path)
+    # Getting the base64 string
+    base64_image = encode_image(image_path)
+    os.remove(image_path)
+
+    headers = {
+      "Content-Type": "application/json",
+      "Authorization": f"Bearer {openai_api_token}"
+    }
+
+    payload = {
+      "model": "gpt-4-vision-preview",
+      "messages": [
+        {
+          "role": "user",
+          "content": [
+            {
+              "type": "text",
+              "text": """
+                  Your role is to identify scientific names of species from zoo signs in images, focusing strictly on extracting the scientific name.
+                  If the image is low quality or unreadable, the response in the JSON will be 'low quality image'.
+                  If no informational sign is detected, it will respond with 'no sign found'.
+                  When multiple signs are present, the response will be 'more than one sign'.
+                  The GPT interacts minimally, responding in a dictionary format with the key "result" and the value being the scientific name or the specific response based on the image analysis.
+              """  
+            },
+            {
+              "type": "image_url",
+              "image_url": {
+                "url": f"data:image/jpeg;base64,{base64_image}",
+                "detail": "low"
+              }
+            }
+          ]
+        }
+      ],
+      "max_tokens": 300
+    }
+
+    response = requests.post("https://api.openai.com/v1/chat/completions", headers=headers, json=payload)
+
+    result = response.json()['choices'][0]['message']['content']
+
+    json_string = "".join(result.split("\n")[1:-1])
+
+    # Parse the JSON string into a dictionary
+    result_dict = json.loads(json_string)
+
+    return result_dict['result']
+
+def get_information(image):
+    taxonomy = get_taxonomy(image)
+    genus, species = taxonomy.split()[0], taxonomy.split()[1]
+    iucn_data = get_iucn_data(genus, species)
+    information = f"## {taxonomy}"
+
+    if len(list(iucn_data.keys())) > 0:
+        information += f"""
+        ## {iucn_data['common_name']}
+
+        **Conservation status**: {iucn_data['status_category']} ({iucn_data['status_code']}).
+
+        **Threats**: {iucn_data['threats']}.
+
+        **Population**: {iucn_data['population']}.
+        
+        **Habitat**: {iucn_data['habitat']}.
+
+        *For more information, please visit this species page*: {iucn_data['web_link']}
+        """
+    
+    return information
+
+image = gr.Image(label="Image", type='pil')
+output = gr.Markdown()
+
+demo = gr.Interface(
+    fn=get_information,
+    inputs=[image],
+    outputs=output,
+    title="ZooSign Reader",
+    examples=['example_1.png', 'example_2.jpeg'],
+    description="""
+    Introducing **ZooSign Reader**, an innovative application designed to enhance your zoo experience! **ZooSign Reader** allows users to effortlessly upload images of zoo informational signs and receive detailed information about the species mentioned on those signs.
+
+    With **ZooSign Reader**, you no longer need to spend time searching for information about a particular animal or bird species while visiting the zoo. Simply capture an image of the sign using your smartphone camera, or choose an existing image from your gallery, and let **ZooSign Reader** do the rest.
+
+    Using cutting-edge image recognition and natural language processing technologies, **ZooSign Reader** quickly analyzes the uploaded image and extracts the text containing the scientific name. The app then searches the IUCN Redlist's extensive database, which includes a wide range of animals, birds, and reptiles found in zoos worldwide.
+    """
+)
+
+demo.launch()