AI Intelligence Module
The Montek AI Module helps landowners and managers of private land, protected areas, national parks, forests, and wetlands turn field data into useful information.
It combines recognition on field devices with centralized analysis on the management platform. Montek AI supports wildlife and aquatic species identification, human disturbance detection, acoustic recognition, environmental risk assessment, and ecological analysis. Results are connected with locations, observation times, and supporting evidence to help managers assess conditions and prioritize action.
Edge AI
Identify Relevant
Information in the Field
On compatible sensing devices and edge computing equipment, Montek AI performs initial recognition, filters observations, and generates alerts before data reaches the management platform.
This allows relevant detections and evidence to be prioritized for transmission, supporting field operations where network capacity is limited or public network coverage is unavailable.
Platform AI
Connect Records and
Assess Conditions
The management platform processes incoming imagery, audio, environmental measurements, and patrol records. Centralized analysis combines detection results with historical events, spatial information, and operational records.
Managers can review classifications, examine evidence, analyze ecological patterns, and assess risks with support from human review and correction.
Six Core AI Capabilities
for Recognition and Risk Analysis
Analyze camera-trap photographs, video, and drone imagery to identify terrestrial wildlife. Results can include species labels, confidence scores, observation times, locations, and supporting imagery, helping teams process monitoring records and examine wildlife activity.
Identify people, vehicles, and potentially concerning activity in camera and drone imagery. Detection results, image excerpts, and alerts help managers review possible unauthorized entry, assess disturbance, and coordinate field verification.
Analyze animal calls, environmental recordings, and potential disturbance sounds, including logging sounds or gunshots. Results connect sound categories with recording times and monitoring locations to support wildlife monitoring, soundscape analysis, and risk alerts.
Analyze underwater photographs and video to identify fish and other aquatic organisms. Classification results, observation frequency, and time records support aquatic monitoring and ecological assessment in rivers, lakes, and wetlands.
Analyze thermal imagery to locate people, animals, and unusual heat sources during night-time or low-light observations. Target locations, image evidence, and observation times support patrols, incident verification, and emergency assessment.
nalyze rainfall, water levels, weather, and soil measurements to identify trends and conditions that may require attention. Threshold alerts and risk assessments support the review of flood conditions, extreme weather, and other environmental hazards.
Utilizes state-of-the-art algorithms to enable high-accuracy,
real-time species identification.
Deep Learning
Multi-species Recognition
Based on long-term data collected from Taohuayuan protected areas, the current models support the identification of 36+ wildlife species,
including rare species such as giant panda, golden snub-nosed monkey, and Sichuan partridge.
The system achieves an identification accuracy of 95% and a recall rate of 93%.
Rapid Annotation and Training
Built on a dataset of 5.5 million real infrared camera images,
MONTEK employs an automated annotation and staged training workflow,
enabling large-scale labeling and model training to be completed within four months.
Technical Advantages
95%
Identification Accuracy
High-precision recognition performance validated through extensive field testing in real protected-area scenarios.
36+
Protected Species Supported
The current models cover over 36 species commonly monitored in nature reserves, including key protected and flagship species.
~100 ms
Response Time
Millisecond-level processing speed enables near real-time detection and monitoring.
