FlyRLECCL

C99 RLE based Connected Component labeling & analysis Library

5-6× Faster Component Statistics Extraction than OpenCV.

Zero Allocations. Zero Dependencies. Complete & Flexible Determinism.

FlyRLECCL is a production-ready, dependency-free C99 SDK engineered specifically for mission-critical embedded vision, smart cameras, medical imaging, and high-throughput document processing. It replaces bloated, non-deterministic frameworks with a hyper-optimized, single-pass RLE engine.

The Core Value: Why Technical Buyers Choose FlyRLECCL

When integrating Connected Component Analysis (CCA) into real-time production systems, modern engineering teams face a strict trade-off between throughput and memory reliability. FlyRLECCL eliminates this compromise.

Hardcore Performance vs. OpenCV (WithStats)

Tested on industry-standard benchmarks (YACCLAB and Tobacco800 datasets), FlyRLECCL drastically reduces CPU cycle consumption:

Component Statistics Are Customizable and Virtually Free

In standard computer vision workflows, enabling geometric statistics often introduces a severe performance penalty. For instance, moving from pure labeling to feature extraction in OpenCV results in a massive throughput degradation on realistic production workloads. Benchmark data reveals a 191.5% overhead on the Medical dataset (from YACCLAB) and a 221.0% overhead on the Tobacco800 dataset just to compute basic component metrics—a systematic behavior widely documented by developers in standard profiling scenarios (e.g., Stack Overflow #52260829)

FlyRLECCL completely rearchitects this step. By computing topological and spatial metrics natively at the interval (RLE) level during the core extraction phase, the computational "tax" is drastically minimized.

The Result: Moving from pure labeling to feature extraction introduces a negligible overhead of only:

FlyRLECCL delivers a comprehensive analytical dataset out-of-the-box, turning what is traditionally a heavy secondary computing step into an immediate, zero-allocation asset:

Architectural Comparison: FlyRLECCL vs. OpenCV

Feature OpenCV
(cv2::connectedComponentsWithStats)
FlyRLECCL
(with Standart Statistic Layout)
Execution Pattern Two-pass pixel dense scanning Two-pass RLE merge
Data Layout of Labeling Results Dense Array of Labels Array of Structures (labeled runs)
Dynamic Allocations Yes Strictly Zero (Pre-allocated buffers)
Memory Fragmentation Risk High (Unsuitable for critical embedded) None (100% Deterministic)
Overhead on Standart Statistics Severe performance drop (200-300%) Minimal performance drop (6% – 14%)
Native Count Metric No (Built-in Parameter - Amount of Runs)

Low-Level Optimization & Hardware Efficiency

Predictive Memory Scaling & "Resume Mode"

Unlike video frames or image matrices, where dimensions are known and fixed in advance, the number of connected components is entirely unpredictable prior to execution. This introduces a critical architectural problem for embedded and real-time systems, forcing developers into a destructive trade-off.

The Static Allocation Dilemma:Standard Computer Vision libraries force developers to budget memory for the catastrophic worst-case scenario (e.g., maximum noise), permanently locking down vast amounts of RAM that go unused 99% of the time.

The Static Allocation Dilemma

To maintain deterministic execution without runtime failures, developers are locked into a binary choice:

The FlyRLECCL Solution: Controlled Budgets and Actionable Telemetry

FlyRLECCL completely resolves this dilemma by pairing a strict zero-allocation model with a specialized Resume Mode for component statistics collection. Instead of budgeting for impossible edge cases, system architects can aggressively minimize the active memory footprint by allocating a micro-buffer scaled to a tight, realistic confidence interval of expected components.

If scene complexity unexpectedly spikes and the number of objects exceeds your allocated boundary, FlyRLECCL safely halts processing and instantly returns FLY_STATUS_STATS_BUFFER_TOO_SMALL. Crucially, even when halting, the engine still writes the exact number of detected components to your n_components pointer. This transforms what is traditionally a fatal application crash into clean, actionable telemetry.

Advanced Pipeline Strategies via Resume Mode

Because n_components is always populated, system engineers can implement highly flexible, adaptive memory and processing strategies inside the hot loop:

By exposing these granular states, Resume Mode provides system engineers with the precise telemetry needed to build robust, flexible, adaptive, and highly deterministic processing pipelines without relying on rigid worst-case constraints or reallocation inside the hot loop.

Integration-Ready: Simple C99 API

Integrating FlyRLECCL into your proprietary software or hardware pipeline takes less than an hour:

Simultaneous Multi-Module Linking (No Conflicts)

Project Roadmap

Commercial Licensing & Integration

FlyRLECCL is distributed as optimized, pre-compiled binary modules (available for major architectures and embedded toolchains, including x86_64, ARM Cortex-A/M, and specialized DSPs).

Every hardware production scale, budgeting cycle, and compliance framework is unique to every enterprise, so to seamlessly align with your business logic, we offer flexible, tailor-made licensing structures instead of rigid, one-size-fits-all contracts.

Available Licensing Frameworks

Comprehensive Integration Support

Every commercial agreement is backed by engineering-focused support to ensure frictionless deployment:

Let’s build the right framework for your pipeline:

We adapt our agreements to your exact deployment scope—whether you are launching a single highly-specialized medical device line or deploying cross-platform smart camera arrays. Contact our licensing team to discuss your operational constraints, request a customized price quote, or align on a contract structure that fits your business model.

Evaluate FlyRLECCL on Your Hardware

See the exact speedup on your target platform before writing a single line of code. We provide a lightweight, dependency-free CLI demonstration engine. Run it against your proprietary datasets and benchmark it side-by-side with your current pipeline.

1. Request the Evaluation Binary or Static Module

To test the core engine performance on your specific architecture (x86, ARM Cortex, or embedded RTOS), you can request a compiled CLI demo executable or a static library module under a time-limited Trial Evaluation License to benchmark it directly inside your application.

2. Integration & Technical Validation

Already running the demo or need to map FlyRLECCL into your existing data pipeline? Connect directly with the core development architecture team to resolve alignment contracts, custom memory padding, or boundary behaviors.

3. Commercial Licensing

Ready to deploy FlyRLECCL into production? Request production-ready SDK binary access tailored to your hardware target, get volume pricing structures, or discuss the best contract alignment for your business model—whether you require our B2B Royalty-Free Licensing for unlimited deployment, volume-scaled Per-Device Licensing, or another Custom/Hybrid Licensing Models.