Hachi

Hachi builds upon Raman Labs' computer vision foundation, combining it with natural language processing for intuitive multimedia search. As a self-hosted solution with local processing capabilities, it offers robust functionality through efficient computational design while maintaining strong security standards.

General Overview

Hachi builds upon Raman Labs' robust computer vision foundation, combining it with natural language processing to create an intuitive multimedia search experience. As a self-hosted web application, it operates entirely offline once configured, offloading all processing to the local machine.

Key technical specifications include a minimum requirement of a quad-core CPU with AVX2 instructions and 4GB RAM, with support for x86_64 architectures. This setup allows the tool to run efficiently on consumer-grade hardware while supporting both individual use cases and small-scale deployment scenarios like virtual private servers (VPS).

The platform covers a comprehensive range of functionalities through several specialized modules. For video content, users can search for specific scenes, objects, or persons while the video plays, thanks to advanced indexing capabilities that process and store metadata locally. Image searches are facilitated through text queries, enabling users to find relevant photos based on keywords or descriptions.

A notable feature is Hachi's face recognition module, which allows users to search for specific individuals or loved ones across their image collections using just a face snapshot. This capability functions entirely locally, bypassing concerns about data privacy and transmission security. All indexed data remains stored on the user's device, further emphasizing the platform's commitment to local processing and privacy.

The development philosophy underlying Hachi emphasizes minimalistic API design and efficient computational implementation. The tool achieves real-time performance even with high-resolution data on consumer-grade CPUs, while its architecture automatically adapts to varying system capabilities. This performance optimization is crucial for ensuring reliable operation across different hardware configurations commonly found in end-user environments.

Core Features and Functionality

Hachi employs a modular architecture derived from Raman Labs' software development kit (SDK), which enables rapid integration of computer-vision capabilities into existing applications through just two lines of code. The system demonstrates remarkable versatility, running on consumer-grade hardware including Windows and Linux systems, and supports both CPU and GPU processing for handling larger workloads.

The video search functionality achieves real-time performance even with high-resolution inputs, thanks to optimized algorithms that maintain efficiency on typical consumer hardware. Users can perform searches while videos play, thanks to advanced indexing that processes and stores metadata locally on the user's device. Similarly, the image search feature allows users to conduct text-based queries across their image collections, drawing upon powerful text-vision models developed by Raman Labs.

Face recognition stands out as a particularly intuitive feature, enabling users to catalog and search their image collections based on facial features alone. Working entirely offline and locally, this module allows users to tag and retrieve photos containing specific individuals or loved ones with remarkable speed and accuracy - a capability made possible through Raman Labs' robust computer-vision technology.

All data storage and processing occurs on the local device, ensuring minimal privacy concerns and maximum responsiveness during searches. The core infrastructure, including email and password storage for registered users, has been designed with security in mind, with payment information processed through Level 1 PCI DSS compliant Razorpay to protect user credentials. This local-first approach combines powerful computational capabilities with stringent privacy standards to deliver a highly functional multimedia search platform that performs seamlessly on a wide range of hardware configurations.

Technical Specifications and Requirements

The tool demonstrates impressive versatility in its implementation, operating on both VPS systems and local installations while requiring only Python3 and NumPy for deployment. Under the hood, it utilizes Japan's Raman Labs SDK, which scales its performance based on the underlying system's capabilities, supporting both CPU and GPU processing for heavier workloads.

Hachi's local-first architecture means it requires minimal system resources beyond the basic specifications - a quad-core CPU with AVX2 instructions and 4GB RAM. This setup allows it to run efficiently on standard consumer hardware while supporting more demanding use cases like virtual private servers. The system automatically adapts its processing power to match the available resources, maintaining optimal performance across different hardware configurations.

From a development perspective, the toolkit emphasizes simplicity and efficiency, allowing integration into existing applications with just two lines of code. This minimalistic API design, combined with powerful computer-vision algorithms, enables real-time performance even with high-resolution data inputs. The cross-platform compatibility, which extends to both Windows and Linux systems, makes it accessible for a wide range of deployment scenarios.

Data Handling and Security

User data management follows a principle of minimalism, storing only essential information such as email addresses and hashed passwords for authenticated users. This approach ensures that personal data remains under direct user control while enabling basic account management features.

Payment processing operates through PCI DSS compliant Razorpay, ensuring that sensitive financial information remains secure throughout the transaction process. No credit card or PayPal details are stored on Hachi's servers, maintaining an additional layer of security for user information.

The development team prioritized simplicity and efficiency in their design choices, implementing the core functionality through just two lines of code using Raman Labs' SDK. This minimalistic API design extends to data handling, where the system requires minimal input to begin operation - users need only provide personal email addresses for account creation and login purposes.

All data processing occurs locally on the user's device, with the system storing indexed content and metadata directly on the local file system. This architecture ensures that even large datasets remain efficiently manageable, with the platform capable of indexing and searching through extensive video and image collections with minimal computational overhead.

Development and Integration

The development of Hachi builds upon Raman Labs' open-source computer-vision software development kit (SDK), which enables seamless integration of advanced ML functionality into existing applications through just two lines of code. The system demonstrates remarkable versatility, operating on both virtual private servers (VPS) and local installations while requiring only Python3 and NumPy for deployment.

The SDK scales its performance based on the underlying system's capabilities, supporting both CPU and GPU processing for heavier workloads. This flexibility allows Hachi to maintain real-time performance even with high-resolution data inputs on typical consumer hardware. The system's cross-platform compatibility extends to both Windows and Linux systems, making it accessible for a wide range of deployment scenarios.

From a development perspective, the toolkit emphasizes simplicity and efficiency. The minimalistic API design enables integration into existing applications with just two lines of code, while powerful computer-vision algorithms maintain optimal performance across different hardware configurations. This architectural approach ensures reliable operation across various deployment scenarios, from small-scale personal usage to larger organizational implementations.

Raman Labs processes payment information through Level 1 PCI DSS compliant Razorpay, maintaining strict security standards for credit card and PayPal transactions. The SDK requires minimal user input, collecting only personal email addresses and hashed passwords for account creation and login purposes. All other data processing occurs locally on the user's device, with the platform storing indexed content and metadata directly on the local file system. This architecture enables efficient management of large datasets while maintaining robust performance and security standards.