🧠 Brain-Inspired Perception Information Processing System

Author: Yajing Zheng
Date: April 2017

📄 This repository contains the implementation code for our paper:
"Spike-Based Motion Estimation for Object Tracking Through Bio-Inspired Unsupervised Learning"
📰 Published in IEEE Transactions on Circuits and Systems I: Regular Papers
🔗 Read the paper on IEEE Xplore

---

⚠️ Quick Note

Since Step 1 is time- and memory-consuming, we have saved the execution results in DATA.mat.
The true innovation of this project lies in Step 2, so you can directly load the DATA.mat file in MATLAB to proceed with the computations.

---

🔬 1. Unsupervised Training with STDP-based HMAX

⚠️ The STDP training files are too large to upload to this repository directly.
After cloning this project, please manually download the STDP-related code and files to the main project directory from the following Baidu Netdisk link:

📦 STDP Files Download:
🔗 https://pan.baidu.com/s/142AA_z4AHyVp-JURZaKYrw?pwd=0601
🔑 Extraction Code: 0601
(Shared via Baidu Netdisk Super Member v2)

1. Run the script stdpRum.m located in the STDP/script directory to perform unsupervised training on the 3D dataset and extract image features.

   After the training, save all feature data into DATA.mat:

   save 'DATA.mat'

---

🧠 2. Encoding and Supervised Training

1. Load the pre-saved data into the workspace:

   load 'DATA.mat'

2. Run OBJREGmain.m for training and evaluation:

   • arithEnc: Arithmetic-based encoding 🧮
     Inputs: COMMON.firingTime, COMMON.localFiringSpike, COMMON.firingSpike, and COMMON.picScale.
     Output: PtnTrSet (spatiotemporal patterns for training) and TrainLabels.

   • TrainSnglN: Trains a weight matrix for each object. 💪
     After training, the weight matrix is stored.

   • Tesing: Uses trained weights to classify test images. 🧪
     Outputs:
     TeFiring (output spike train),
     DistTePtns (vR distance to target spike train).

   • ResultAnalysis: Evaluates recognition results based on DistTePtns. 📊

---

🙌 Support & Citation

If you find this project helpful, please consider giving it a ⭐️ or citing our paper to support our research!
Your support encourages us to continue exploring brain-inspired AI. 🌟🧠

📚 Cite our work:

@article{zheng2022spike,
  title={Spike-Based Motion Estimation for Object Tracking Through Bio-Inspired Unsupervised Learning},
  author={Zheng, Yajing and others},
  journal={IEEE Transactions on Circuits and Systems I: Regular Papers},
  year={2022},
  publisher={IEEE}
}

🔗 https://ieeexplore.ieee.org/abstract/document/9985998