06 · Edge AI · Computer vision · Geospatial · Research
Real-Time Wildlife Monitoring
Research paper: “Adaptive Occlusion Edge Intelligence for Real-Time Wildlife Monitoring and Water Source Analysis Across Diverse Ecosystems” - Mohit Patel, Jabalpur Engineering College
An edge-AI system on sub-$150 hardware that detects, tracks and maps wildlife around water sources - 87.3% mAP@0.5 at 7.7 FPS on a Raspberry Pi 4, with an occlusion-aware method that adds 11.2 points under forest canopy.
87.3%
mAP@0.5 (15 species)
7.7 FPS
Raspberry Pi 4 · 5.9 W
+11.2 pts
Under heavy canopy
The problem: water sources decide where wildlife gathers, especially in summer, but collars, camera traps and manual surveys are slow, costly and intrusive.
Ground nodes: a Raspberry Pi 4 with Camera Module v2 and a Neo-6M GPS (5-8 m accuracy) runs the full pipeline at the water hole; ESP32-CAM units wake on a PIR sensor and take one image every 5 s at 0.8 W.
Aerial survey: drone imagery over a 2 km² mixed-deciduous forest plot in central India - 50 m altitude was chosen after testing 30, 50 and 70 m. My own 2.5-inch FPV drone carries a Raspberry Pi and a detection camera for low passes over water holes.
Detection: YOLOv8n fine-tuned on 12,000 hand-annotated frames of 15 species (forest plot, drone and Snapshot Serengeti), quantised to INT8 - 89 ms per frame on the Pi.
My occlusion-aware confidence modulation scores how hidden an animal is from image texture and edges, and boosts confidence near water: under heavy canopy, mAP rose from 75.8% to 87.0%.
Tracking: Deep SORT with an occlusion penalty keeps each animal’s ID - 55% fewer ID switches (4.1 vs 9.2 per 1,000 frames) and MOTA up from 73.8% to 79.1%.
Mapping: every detection is GPS-tagged and pinned on the map; kernel-density heatmaps with 50/75/90/95th-percentile contours reveal saturation zones - 124 m mean radius around the main water source, from 142 m for chital to 85 m for leopard.
Counting: animals are tallied per species and saved for the forest administration, so they know how many of each species use each water source.
Results: 87.3% mAP@0.5 overall, 5.7 points above MegaDetector v5; 90.4% species classification; whole pipeline 130 ms per frame (7.7 FPS) at 5.9 W.
Custom model for university student timetable scheduling
A genetic algorithm model that builds university student timetables, with role-based functionality - written in Python.
GA
Genetic algorithm core
Python
Custom model
Roles
Role-based functionality
Requirement: build a custom model for university student timetable scheduling, with role-based functionality.
Approach: a genetic algorithm - a population of candidate timetables is scored, the fittest are combined (crossover) and randomly tweaked (mutation), generation after generation, until a clash-free timetable emerges.
Role-based functionality: different university users get the features that match their role.
Phase 1 built: the web app and its models · Phase 2 planned: a wearable locket
An AI voice-analysis web app that listens 24×7, picks out what matters in your conversations - meetings, who said what, tasks - puts events and tasks into Google Calendar, sends voice reminders and gives you a summary at the end of the day.
24×7
Always listening
Calendar
Events & tasks auto-added
Daily
End-of-day summary
Listens to your voice 24×7 and records the data.
When something important comes up - like a meeting in that day’s conversation - it captures what was talked about, with whom, and the details.
Scheduled tasks and events are added to Google Calendar automatically.
Voice reminder alerts before things are due, and a summary of your day at the end of it.
Phase 1 (built): the core algorithm with sentiment analysis, speech-to-text and a summary model.
Phase 2 (planned): a hardware locket with a microphone that records messages and does much more on the go.
A front-facing mobile camera that detects road accidents. When a crash happens it sends an SMS and an SOS call to the local police station and the nearest government hospital and shares live GPS location - using acceleration and gyro data so it doesn’t raise false alerts.
SOS
Police + govt hospital
Live GPS
Location shared
Accel + gyro
No false alerts
A mobile phone is mounted with its camera facing the road ahead.
If an accident happens, it sends an SMS and makes an SOS call to emergency services - the local area police station and the nearest government hospital.
Live GPS location is shared so help reaches the exact spot.
It reads all the phone’s sensors - acceleration, gyroscope and more - so bumps and potholes don’t trigger false alerts.