Bharat in Betastartup field guide
← Back to the atlas
No. 1402Problem statementSpace and research

AI Human Activity Recognition for On-board BAS Experiments

Source

Source summary

Train a standalone onboard activity-recognition assistant for predefined BAS experiments using fixed cameras and locally generated training data. Track steps, suggest the next action and issue voice alerts for skipped/out-of-order actions. Produce lightweight timestamped outcome/status logs, store video locally and support streaming to a specified IP, with a GUI. Orientation-agnostic 3D body modelling is optional. The model must run offline; flawless experiment execution is not guaranteed by activity recognition.

No. 1291Problem statementSpace and research

Development of a Low-Power, Real-Time Adaptive Software-Defined Sonar Transmitter Payload for Autonomous Underwater Vehicles (AUVs)

The requested AUV sonar payload must be a physical, low-power embedded transmitter prototype rather than simulation alone. Environmental sensor inputs, or simulated sensor controls, should adapt the transmitted waveform in real time. The source asks for multiple modulation formats, hardware-efficient streaming, signal conditioning, an enclosure and instrument-based output validation. This entry retains the marine mapping hardware scope without turning it into a circuit-construction tutorial.

No. 1046Problem statementSpace and research

Target detection by optimizing Anomaly Detection in Hyperspectral Image Processing using AI/ML

Research AI/ML anomaly detection in public hyperspectral datasets such as Hyperion or AVIRIS, identifying spectrally unusual pixels and object signatures. The source discusses preprocessing, denoising, fusion and radiometric/atmospheric correction before evaluating deep-learning models. This overview retains image-analysis research scope without target-engagement instructions; improved detection remains to be demonstrated.

No. 1112Problem statementSpace and research

Integrated Geo-Referenced Fish Catch Data Repository and Access System

Create an INCOIS repository consolidating georeferenced fish-catch records from researchers, industry, cruises and image-based collection. Accept Excel data and organise by date, location, depth and species, with filters, visualisation and downloads for species-specific habitat research. Preserve distinctions between species-level weights and aggregate catch weights with species presence. Use separate occurrence and abundance modules, geographic/time filters, mapped results, Excel/CSV exports and authenticated access with an updating administration interface.