Hi, I'm Aafaq Inamdar,
Computer Vision Researcher implementing AI for making human life better!
Hi, I'm Aafaq Inamdar,
Computer Vision Researcher implementing AI for making human life better!
📍 Mumbai, India.
📍 Mumbai, India.
Embedded Files
Computer Vision Researcher implementing AI for making human life better!
Real-time computer vision that’s accurate, fast, and production-ready.
Computer Vision | Deep Learning | Sports Science | Synthetic Data
Real-time computer vision that’s accurate, fast, and production-ready.
Computer Vision | Deep Learning | Sports Science | Synthetic Data
About me
About me
I’m a Computer Vision & Deep Learning Engineer focused on real-time perception, 3D calibration/triangulation, and GPU-accelerated optimization.
Currently am working at Quidich Innovation Labs, I lead vision initiatives powering broadcast-grade ball tracking and calibration across 4K, High FPS, multi-camera setups—using TensorRT, CUDA, and NVIDIA DALI to hit sub-ms sync and dramatically reduce trajectory error. Also, image stabilisation methods to preseve the calibration in outdoor sports environment when the cameras are zoomed in.
I also build surveillance and smart-city solutions (action recognition, abandoned-object detection, traffic analytics, ID/OCR), synthetic data pipelines (Unity/Unreal/Blender/Omniverse), and deploy on Jetson and cloud with rigorous performance tuning.
What I BuildSports & BroadcastMulti-camera ball tracking with millimeter-level triangulationOn-ground calibration (total-station + bundle adjustment)Real-time scene classification for production workflows
Surveillance & Smart CityAbandoned object and perimeter-intrusion detectionTraffic density, vehicle counting, and signal timing analyticsID/OCR (PAN/Aadhaar/Passport MRZ) with CRAFT/Tesseract/PaddleOCRHuman activity recognition with pose estimation + ConvLSTM
Industrial & RetailInventory/security monitoring on Jetson (edge)Low-latency detection/tracking for shop-floor & logistics
Synthetic Data & ResearchProcedural synthetic datasets (Unity, Unreal, Blender, Omniverse)Domain gap reduction, model distillation, diffusion-based augmentationOptimization & DeploymentTensorRT quantization, CUDA kernels, NVIDIA DALI pipelinesEdge (Jetson/RPi) to multi-GPU cloud orchestrationC++/Python, PyTorch/OpenCV, GPU-Direct ingestion
Surveillance & Smart CityAbandoned object and perimeter-intrusion detectionTraffic density, vehicle counting, and signal timing analyticsID/OCR (PAN/Aadhaar/Passport MRZ) with CRAFT/Tesseract/PaddleOCRHuman activity recognition with pose estimation + ConvLSTM
Industrial & RetailInventory/security monitoring on Jetson (edge)Low-latency detection/tracking for shop-floor & logistics
Synthetic Data & ResearchProcedural synthetic datasets (Unity, Unreal, Blender, Omniverse)Domain gap reduction, model distillation, diffusion-based augmentationOptimization & DeploymentTensorRT quantization, CUDA kernels, NVIDIA DALI pipelinesEdge (Jetson/RPi) to multi-GPU cloud orchestrationC++/Python, PyTorch/OpenCV, GPU-Direct ingestion
The projects below showcase some of my university and early research work. Most of my recent projects are either featured on company websites or remain confidential due to their commercial or surveillance nature. This is a brief overview, not an exhaustive list.
The projects below showcase some of my university and early research work. Most of my recent projects are either featured on company websites or remain confidential due to their commercial or surveillance nature. This is a brief overview, not an exhaustive list.
Object detection on Nvidia GPU
Object detection on Nvidia GPU
Human pose estimation on stereo oak d pro VPU
Human pose estimation on stereo oak d pro VPU
IOT intrusion detection
Demo!
Page updated
Google Sites
Report abuse