Data

Tools

Indexes

Applications

Experiments

Awesome

An open API service indexing awesome lists of open source software.

https://github.com/sayandeepmaity/luminator

Microphone Array-Based Direction of Arrival of Gunshot Detection .Gun violence remains a critical concern. Identifying the precise location of a gunshot—or getting as close as humanly possible—is crucial for saving lives and ensuring public safety.
https://github.com/sayandeepmaity/luminator

classification cnn cst deep-learning fpga gelu gunshot-detection neural-network pandas prediction pytorch regression rnn skicit-learn tdoa transformers verilog-hdl

Last synced: 25 days ago
JSON representation

Microphone Array-Based Direction of Arrival of Gunshot Detection .Gun violence remains a critical concern. Identifying the precise location of a gunshot—or getting as close as humanly possible—is crucial for saving lives and ensuring public safety.

Awesome Lists containing this project

README 

          
# LUMINATOR

**LIGHTWEIGHT ULTRA-ADVANCED MULTI-SENSOR INTELLIGENT NOISE-ABRIDGE THREAT OVERCOMING RADAR**



---



## Problem Statement Title

**MICROPHONE ARRAY-BASED DIRECTION OF ARRIVAL OF GUNSHOT DETECTION**



It's important to know the **precise location** of a gunshot — or at least as close as humanly possible — because **gun violence** isn't going away anytime soon.  

Our goal is to **harness any available technology** or assisting capability to ensure the **best possible outcome in worst-case scenarios.**



---



## Proposed Solution



An **advanced gunshot detection and classification system** using a **hybrid approach** of:

- Microphone arrays

- FPGA-based signal processing

- Deep learning models



By leveraging **CNN-RNN neural architectures** and **Transformer layers**, we enable:

- Real-time gunshot identification

- 3D localization

- Type classification



---



## How It Solves the Problem



- Instant feedback with highly accurate 3D localization using TDOA

- Multi-classification approach to detect and identify firearm types in real-world scenarios



---



## Role of Deep Learning



- Distinguishes gunshots from other loud noises or distractions

- Reduces false alarms

- Ensures fast & accurate responses



---



### Key Features:

- Immediate detection & 3D localization

- FPGA-based noise filtering and feature extraction

- Deep learning-based classification

- Real-time responsiveness



---



## Innovation and Uniqueness



- Hybridization: Combining CNNs and RNNs for enhanced audio analysis

- Transformer Integration: Improved attention to relevant features

- Real-time Application: Live analysis of hazardous sound events



---



## Technologies Involved



### Programming Languages

- Python (ML Models)

- Verilog (FPGA Processing)



### Hardware

- 4 or 6 omnidirectional microphones

- FPGA for real-time signal processing

- GPU/TPU for model training

- LCD Display for output



### Frameworks & Libraries

- TensorFlow

- PyTorch

- Scikit-Learn

- Pandas

- TQDM



### Machine Learning Models

- CNN

- RNN (LSTM)

- Regression (TDOA)

- CNN + Fully Connected

- CST Transformer



---



## Methodology & Implementation



1. Microphone Array captures the sound

2. ADC converts it to digital signal

3. FPGA applies bandpass filtering (up to 3kHz) & extracts features

4. CNN-RNN Hybrid Model processes features

5. CST Transformer Layer applies 3 attention types:

    - Channel-MHSA: Spatial attention

    - Spectral-MHSA: Frequency-based attention

    - Temporal-MHSA: Time-based evolution attention

6. GELU Activation enhances model performance

7. Real-time output shown on display



---



## Technical Feasibility



- Built on existing technologies like FPGAs and Neural Networks

- Uses real-time signal processing and deep learning

- Feasible and scalable with current hardware



---



## Why CNN + RNN Hybrid?



### CNNs:

- Extract local patterns and hierarchies (e.g., sound textures)

- Reduce dimensionality while preserving important spatial info



### RNNs:

- Retain temporal memory

- Model long sequences (like varying gunshot durations)

- LSTM or GRU helps with temporal context in audio



---



## Model Performance Comparison



| Model               | Gunshot Detected (%) | Muzzle Blast Detected (%) | Shockwave Detected (%) | TDOA Accuracy (%) | Gun Type Classification (%) |

|--------------------|----------------------|----------------------------|-------------------------|-------------------|------------------------------|

| CNN-Only           | 65                   | 83                         | 89                      | 72                | 80                           |

| RNN-Only           | 83                   | 85                         | 94                      | 79                | 91                           |

| DNN-Only           | 82                   | 90                         | 96                      | 77                | 93                           |

| CNN+RNN            | 92                   | 93                         | 98                      | 82                | 95                           |

| CNN+RNN+CST        | 96                   | 95                         | 97                      | 89                | 98                           |



---



## Challenges & Risks



- Hardware Limitations: FPGAs have finite resources

- Latency: Real-time needs demand high processing speed

- Noise Interference: May affect accuracy

- Data Availability: High-quality gunshot datasets are rare



---



## Optimizations & Techniques



### FPGA Resource Optimization:

- Designed efficient Verilog modules

- Implemented parallel processing for real-time handling



### Algorithm Improvements:

- Developed real-time, low-latency algorithms for gunshot detection

- Integrated high-pass/low-pass filters

- Trained ML models with noise-augmented datasets



---



## References



- [FPGA Filters](https://ashrafi.sdsu.edu/PDF/filters_FPGA.pdf)  

- [Gunshot Detection Thesis - Auraria Library](https://digital.auraria.edu/files/pdf?fileid=e562b890-41f9-4170-a13b-27b07d1f1626)  

- [TDOA Localization Field Guide](https://www.decodio.com/media/downloads/TDOA_Localization_From_Theory_to_the_Field_v1-0.pdf)  

- [Multiple Impulse Acoustic Sources - MDPI](https://www.mdpi.com)  

- [Gunshot Detection Using Accelerometers - PLOS ONE](https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0199023)  

- [Gun Identification using Transformer - Nature Scientific Reports](https://www.nature.com/articles/s41598-021-88594-3)  

- [Raytheon's Boomerang Acoustic System](https://prd-sc102-cdn.rtx.com/-/media/rtx/our-company/who-we-are/bbn/boomerang/files/boomerang_data.pdf)  

- [Object Tracking using TDOA - MATLAB](https://www.mathworks.com/help/dsp/ug/object-tracking-using-time-difference-of-arrival.html)  

- [Gunshot-like Sounds Detection - MDPI](https://www.mdpi.com)  

- [Postprint PDF - Google Drive](https://drive.google.com/file/d/1g3ceUoOC9cOvfQ/pdf)  

- [FireBrick Project - Tufts University](https://sites.tufts.edu/eeseniordesignhandbook/files/2017/05/FireBrick_OKeefe_F1.pdf)  

- [Indoor Gunshot Notification System - MDPI](https://www.mdpi.com)