deploy

.gitignore

@@ -0,0 +1 @@
+.env

README.md

@@ -1,6 +1,6 @@
 ---
 title: Planthy
-emoji: 😻
+emoji: 🌿
 colorFrom: gray
 colorTo: pink
 sdk: gradio
@@ -8,7 +8,39 @@ sdk_version: 5.33.0
 app_file: app.py
 pinned: false
 license: apache-2.0
-short_description: a plant health monitoring agent
+short_description: a plant health monitoring app
+tags:
+  - agent-demo-track
+
 ---
 
-Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
+# Planthy -- Plant Health Monitor Agent
+
+## Description
+
+A multimodal AI agent that diagnoses plant health issues by analyzing images and text queries, providing actionable recommendations. Built for a hackathon, it leverages real-time web search and advanced reasoning to assist gardeners and farmers, deployed as a user-friendly web app on Hugging Face Spaces.
+
+## Tech Stack
+
+* LlamaIndex: ReAct agent and tool orchestration.
+
+* Gemini: Multimodal image and text processing (`gemini-2.0-flash`).
+
+* DuckDuckGo Search: Real-time plant health information retrieval.
+
+* Gradio: Web interface for user interaction.
+
+## Functionality
+
+* Image Analysis: Upload a plant image to identify type, condition, symptoms, and confidence (e.g., "Tomato, Diseased, Yellow spots, 0.85").
+
+* Text Queries: Input queries (e.g., "What’s wrong with my plant?") to get tailored recommendations.
+
+* ReAct Reasoning: Combines image analysis and web search for intelligent, context-aware responses.
+
+* Web Interface: Gradio UI for easy image uploads, query input, and Markdown-formatted outputs (diagnosis, recommendations, reasoning).
+
+* Real-Time Data: Fetches plant care tips via DuckDuckGo search, eliminating the need for a local dataset.
+
+## Demo
+

app.py

@@ -0,0 +1,97 @@
+import gradio as gr
+from PIL import Image
+import os
+from dotenv import load_dotenv
+from llama_index.multi_modal_llms.gemini import GeminiMultiModal
+from llama_index.core.program import MultiModalLLMCompletionProgram
+from pydantic import BaseModel
+from llama_index.core.output_parsers import PydanticOutputParser
+from llama_index.core.tools import FunctionTool
+from llama_index.core import SimpleDirectoryReader
+from duckduckgo_search import DDGS
+from llama_index.core.agent import ReActAgent
+from llama_index.llms.gemini import Gemini
+
+
+load_dotenv()
+
+token = os.getenv("GEMINI_API_KEY")
+
+class PlantHealth(BaseModel):
+    plant_type: str
+    condition: str
+    symptoms: str
+    confidence: float
+
+prompt_template_str = """Analyze the plant image and return:
+- Plant type (e.g., tomato, rose)
+- Condition (e.g., healthy, diseased)
+- Symptoms (e.g., yellow spots, wilting)
+- Confidence score (0.0 to 1.0)
+"""
+
+def analyze_plant_image(image_path: str) -> dict:
+    gemini_llm = GeminiMultiModal(model_name="gemini-2.0-flash")
+    llm_program = MultiModalLLMCompletionProgram.from_defaults(
+        output_parser=PydanticOutputParser(PlantHealth),
+        image_documents=[SimpleDirectoryReader(input_files=[image_path]).load_data()[0]],
+        prompt_template_str=prompt_template_str,
+        multi_modal_llm=gemini_llm,
+    )
+    return llm_program().dict()
+
+image_tool = FunctionTool.from_defaults(
+    fn=analyze_plant_image,
+    name="analyze_plant_image",
+    description="Analyzes a plant image to identify type, condition, and symptoms."
+)
+
+def search_plant_info(query: str) -> str:
+    with DDGS() as ddgs:
+        results = ddgs.text(query, max_results=3)
+        return "\n".join([f"{r['title']}: {r['body']}" for r in results])
+
+search_tool = FunctionTool.from_defaults(
+    fn=search_plant_info,
+    name="search_plant_info",
+    description="Searches the web for plant health and care information."
+)
+
+agent = ReActAgent.from_tools(
+    tools=[image_tool, search_tool],
+    llm=Gemini(model_name="gemini-2.0-flash"),
+    verbose=True
+)
+
+
+def plant_health_app(image, user_query):
+    image_path = "temp_image.jpg"
+    image.save(image_path)
+    
+    # Query the ReAct agent
+    response = agent.chat(f"Analyze this plant image : {image_path}, give a recommendations to cure the issue and return the answer in points. User query: {user_query}")
+    
+    # Parse response (assume agent returns structured text)
+    recommendations = str(response)
+    
+    output_text = recommendations
+    return output_text
+
+iface = gr.Interface(
+    fn=plant_health_app,
+    inputs=[
+        gr.Image(type="pil", label="Upload Plant Image"),
+        gr.Textbox(label="Describe Symptoms or Ask a Question", placeholder="E.g., My tomato plant has yellow spots")
+    ],
+    outputs=gr.Markdown(label="Diagnosis and Recommendations"),
+    title="🌿 Planthy -- Monitor Your Plant Health",
+    description="Upload a clear plant image and describe symptoms to get a diagnosis and care tips.",
+    theme="huggingface",
+    examples=[
+        ["tomato.jpeg", "What’s wrong with my tomato plant?"],
+        ["rose.jpeg", "Is my rose plant healthy?"]
+    ]
+)
+
+if __name__ == "__main__":
+    iface.launch()

requirements.txt

@@ -0,0 +1,8 @@
+llama-index 
+llama-index-multi-modal-llms-gemini 
+llama-index-embeddings-gemini 
+google-generativeai 
+gradio 
+pillow 
+matplotlib 
+duckduckgo-search