Course Objective & Scope
Moving a model that works on single images onto 24/7 camera feeds quickly exposes dropped frames, ID switches, bandwidth bottlenecks and false alarms. Starting from RTSP/WebRTC ingestion, this course covers frame sampling, real-time detection, multi-object tracking (MOT) with ByteTrack and BoT-SORT, and intra/inter-camera Re-ID.
Enterprise scenarios such as line-crossing counts, dwell time, zone intrusion and crowd heatmaps are implemented with GStreamer, Supervision and NVIDIA DeepStream. Participants design a real-time video analytics architecture and event-triggering flow suited to their own field cameras.
Who Should Attend?
- Computer vision engineers building real-time systems on camera and sensor feeds
- Software developers working on smart facility, retail, transport and security projects
- Platform and infrastructure engineers managing high-volume video streams
- Prerequisite: Python and basic object detection knowledge
Key Takeaways
- Ingest and process RTSP camera streams at low latency with FFmpeg and GStreamer
- Maintain consistent object identities under occlusion using ByteTrack and BoT-SORT
- Compute line-crossing, zone intrusion, dwell-time and crowd-density metrics
- Measure tracking quality with HOTA, MOTA and IDF1 metrics
- Scale multi-camera streams on GPUs using batching and DeepStream
Curriculum
Video Streaming Architecture and Decoding
- RTSP, WebRTC, H.264/H.265 codecs, GOP structure and hardware-accelerated decoding (NVDEC)
- Stream ingestion, frame buffering and dynamic sampling with FFmpeg and GStreamer
Multi-Object Tracking (MOT)
- Kalman filtering, IoU matching and the Hungarian algorithm
- ByteTrack, BoT-SORT and appearance-based Re-ID embeddings
Spatial and Temporal Analytics Rules
- Polygon zones, virtual lines, directional counting and speed estimation
- Dwell-time analysis, queue monitoring and heatmap generation
Multi-Camera (MTMC) and Scalable Pipelines
- Camera calibration, homography and cross-camera identity matching
- Multi-stream batching with NVIDIA DeepStream and message broker (Kafka/MQTT) integration
Evaluation, Alert Management and Privacy
- Field performance analysis with HOTA, MOTA and IDF1 metrics
- False-alarm debouncing and automatic face/plate anonymisation
Other Courses in This Category
-
Human action, ergonomics and anomaly detection with 3D CNNs (SlowFast, X3D), VideoMAE/TimeSformer and skeleton-based models (MMPose, ST-GCN), for teams building workplace safety, assembly-line and behaviour analysis systems.
- Live Course
- 5 modules
-
Shared embedding spaces with CLIP/SigLIP, Qwen2.5-VL, LLaVA and InternVL architectures, visual QA, multimodal RAG and QLoRA fine-tuning, for engineers building systems that reason across images and text.
- Live Course
- 5 modules
-
Surface defect detection on production lines, unsupervised anomaly detection with anomalib (PatchCore, PaDiM, EfficientAd), industrial camera/lighting design and PLC integration, for quality and manufacturing engineering teams.
- Live Course
- 5 modules
-
Inpatient fall prevention, ICU monitoring, hand-hygiene and PPE compliance, operating-room instrument counting and privacy-preserving edge anonymisation, for hospital digital transformation and healthtech teams.
- Live Course
- 5 modules
-
DICOM/NIfTI processing, 2D/3D lesion and organ segmentation on CT/MRI/X-ray with MONAI and nnU-Net, whole-slide pathology (WSI) analysis and clinical validation, for medical imaging and biomedical AI teams.
- Live Course
- 5 modules
-
Reception and lobby queue/crowd analytics, buffet tray replenishment and food-waste measurement, car-park ANPR, pool/facility safety and privacy-compliant anonymous analytics, for hospitality technology teams.
- Live Course
- 5 modules
Get in touch
