thermal · RGB · aerial
Small object detection
Deep-learning detectors for targets only a few pixels wide in thermal and RGB aerial imagery, where every pixel of signal counts.
Systems engineer · AI researcher
HaleYıldız
I build deep-learning models that find small objects in thermal and RGB aerial imagery — and make them light enough to run at the edge.
- Co-founderHB Tech
- AI & Sensing LeadŞahit
- PhD studentBaşkent University
About
I’m an electrical and electronics engineer working where systems engineering meets deep learning. I build models that find what is hard to see — people, vehicles, animals and drones only a few pixels wide in thermal and RGB imagery — and recognise them in acoustic signals too. Then I make those models light enough to run on the device itself.
- 3sensing modalitiesThermal, RGB and acoustic
- 5target classes in ŞahitPeople, vehicles, wildlife, livestock, drones
- PhDin progressElectrical & electronics engineering, Başkent University
Focus
What Iwork on
Five threads run through my research and engineering — from a few hot pixels to a fielded system.
params · FLOPs · FPS
Efficient models for the edge
Measuring what a model really costs — parameters, FLOPs, FPS — and adapting lightweight models to run on edge devices.
collect · label · evaluate
Datasets & annotation
Building and labelling field datasets — the unglamorous foundation every reliable detector stands on.
DSP · spectra · features
Signal processing
Digital signal processing for acoustic and sensor data: from raw waveforms to features a model can learn from.
requirements · architecture
Systems engineering
Requirements analysis and system architecture, shaped by systems engineering work in the defense industry.
- Object detection
- Small object detection
- Thermal imagery
- RGB aerial imagery
- Dataset creation
- Data annotation
- Params · FLOPs · FPS
- Edge AI
- Signal processing (DSP)
- Python
- CUDA C/C++
- Machine learning
- Requirements analysis
- System architecture
Featured project
Şahit
an HB Tech projectRoleAI & Sensing Lead
Şahit is HB Tech’s AI-powered sensing network: solar-powered nodes around fields, farms and facilities that tell what is approaching apart — right on the device — and raise the alarm.
My part
- Detection and classification models built on thermal/RGB imagery and acoustic signals, running on the sensor nodes
- Five classes: people, vehicles, wildlife, livestock and drones
- Building and labelling the field datasets
- Optimising the models for edge devices and evaluating them in field tests
- Requirements analysis, system architecture and adapting the platform to defense and critical-facility scenarios
Experience
Where I’veworked
Co-founder
A technology company that designs software and electronics side by side.
hbtech.com.trAI & Sensing Lead
Perception models, field datasets and field-test evaluation for an AI-powered sensing network.
sahit.ioSystems Engineer
Systems engineering in the defense industry: requirements analysis and system architecture.
Candidate Engineer
Candidate engineer programme, before joining STM as a systems engineer.
Candidate Engineer
Candidate engineer programme in communications technology.
Education
Electrical & electronicsall the way through
- PhD2026 – presentDoctorate
Electrical & Electronics Engineering
Başkent University
- MSc2024 – 2026Master’s degree
Electrical & Electronics Engineering
Başkent University
- BSc2017 – 2023Bachelor’s degree
Electrical & Electronics Engineering
TOBB University of Economics and Technology
Certifications
- NVIDIA Deep Learning InstituteFundamentals of Accelerated Computing with CUDA C/C++
- BTK AkademiMachine Learning with Python
- CourseraDigital Signal Processing
Contact
Let’s talk.
Research collaborations, Şahit, or a conversation about seeing more with fewer pixels — my inbox is open.