Abhishek Chakraborty
Computer Vision Engineer | Real-Time Video Analytics | YOLO • RT-DETR • PyTorch • OpenCV | TensorRT • NVIDIA Triton | Edge AI amp; Deep Learning
Greater Hyderabad Area
< 1 year of work experience
Dec 202510 mosSep 2026
Hyderabad
Worked on real-time computer vision and video analytics solutions for airport and aviation operations.• Developed and optimized object detection and tracking pipelines using Python, PyTorch, YOLO, RT-DETR, OpenCV, and ByteTrack.• Worked on passenger and aircraft turnaround analytics, including boarding/deboarding events and real-time video-based monitoring.• Worked with ONNX, TensorRT, and NVIDIA Triton for optimized model inference and deployment.• Worked with edge and GPU-based environments, focusing on real-time inference, performance optimization, and reliable CV pipelines.
Dec 202510 mosSep 2026
Built and improved real-time computer vision systems used for airport turnaround and baggage-handling operations. Built a passenger-counting pipeline using YOLOv7 and ByteTrack, using appearance, movement, and tracking information to distinguish crew from passengers. Fine-tuned and tested YOLOv12-s and RT-DETRv2 models and compared their performance with YOLOv7 using ONNX and TensorRT. Built an optical-flow-based system for baggage loading and unloading detection at airport sites in Lyon. Developed detection logic for aircraft events such as On-Block, Off-Block, and belt-loader activities, using Bi-LSTM and ROI-based methods to improve detection under camera movement and other visual changes. Investigated false positives, missed detections, and unstable confidence scores using system logs and video data; monitored model performance (confidence, latency, resource usage). Configured Milestone VMS for multi-camera video integration in airport projects.
2024Less than a year2024
• Developed a DCGAN-based image inpainting system for image restoration, improving the model through architecture changes and
data augmentation.
• Exported the trained model to ONNX for faster inference and evaluated image quality using SSIM against traditional inpainting
methods.
20221 yr2023
• Built a monkey detection system for Raspberry Pi-compatible hardware: evaluated YOLOv3-Tiny, YOLOv5, EfficientDet-D0 and
EfficientDet-D1, selected EfficientDet-D0, and applied quantization to achieve over 6 FPS on a single thread.
• Collected and annotated images, trained with TensorFlow Model Maker, and evaluated detections using IoU. Developed a modular
model handler and Raspberry Pi 4 integration with camera capture, GPIO, system-health monitoring and Telegram notifications;
used multi-frame aggregation to reduce false positives.
• Developed an active-learning demo using VGG16/ResNet50 to identify uncertain unlabeled images for human annotation and
iterative retraining.
• Built an Aadhaar masking pipeline using PDF-to-image conversion, OpenCV contour detection, CNN-based card detection, OCR,
and blurring of the first eight digits.
2022Less than a year2022
• Deployed YOLO-based object detection on NVIDIA Jetson Nano for real-time applications, reducing false detections by around
15% through model and threshold improvements.
• Developed an automated box detection system for ITC.
Abhishek’s AI Stack
Education
Bachelor of Commerce
The Bhawanipur Education Society College
Join Abhishek and 10k+ other professionals on
Tal is the app where bosses meet engineers & builders directly, without middlemen
Download Tal Boss