EcoCred

• 

  query

• 

  response

Inspiration

Inspired by the pressing need for sustainable solutions, EcoCred emerged with a mission to simplify access to pertinent information for eco-conscious decision-making.

What it does

EcoCred harnesses multimodal data to efficiently search documents, providing pertinent responses that support sustainable practices.

How we built it

Our approach involved: Project creation in Vertex AI Utilizing the generative model Gemini Pro Vision Collection and chunking of PDF data Extraction of image and text metadata Conducting similarity searches by matching user text queries with chunk embeddings Gaining critical insights from both text and images.

Challenges we ran into

Fine-tuning the models to ensure accurate and swift responses posed a significant challenge.

Accomplishments that we're proud of

Despite challenges in fine-tuning models for accuracy and speed, we successfully integrated various data types to develop a robust and responsive search engine for sustainable documents using Gen AI.

What we learned

Through this process, we deepened our understanding of multimodal data processing and optimization, enhancing our proficiency in AI-driven document analysis.

What's next for EcoCred

Looking ahead, we aim to expand EcoCred's capabilities to include real-time carbon credit verification, delve into deeper insights into sustainable practices, and also incorporate video search functionality.

Built With

• geminipro
• google-cloud
• llm
• python
• vertexai