SMRITI-IoT and AI based brain training and cognitive testing system

Discussed with Subodth Sir about the GMCT project
integrated the auto-comport selection feature in web
Adding feature to read and display the spo2 and pulse values.

Transaction of backend Arduino connection logic from backend to frontend.

Implemented HTTPS on the frontend, with which we can connect the Arduino from any pc via the browser but since the back end was on http there was a mismatch so the object detection was not working To fix this I implemented a proxy server, coddy and was able to run the model process in the back end while the detection can be seen in the frontend pc.

Meeting about the project on the anaesthesia depth monitoring system.
fixing the pulse vs time graph plotting issue

Fixed the performance table.
Implementing the export report and print option.

Added the export and report creation in the performance table module.
Added the Arduino access module, license request module, and launch test module and finalizing the user registration module in the Doctors dashboard.

Added the option to scan the QR code and attach a screenshot of the payment in the license module, and the admin can view the screenshots and approve the request.
made a few changes in the proposal.
Successfully learned how to install and set up Nifi.

added the sms sending feature, where when a user submits a licence request to the admin he will also be notified by an email.

Added the touch count per exercise graph, Time taken per exercise graph, speed per exercise graph and Touch frequency distribution graph.
Configured the nifi setup to send the Parquet file to Hadoop.

Changed the report layout as per the client's request
created the installation package for FM-T and FM-E, and shared it with the client.
Meeting discussion related to the project "Big Data Analytics Platform and Tools for Epidemic Forecasting and Mitigation"

Proposal preparation for ICMR

trained and tsted a model that works well in webcam with good precision.
working on lap to fix the lagging issues.

tried to install the smriti.exe in Raspbian but was not able to successfully install the software necessary for running the exe in Linux.
Later tried to run the Python code, but the installation of Torch kept failing.

Worked on the complete proposal submission for the ANRF ‘Maha MedTech Mission’ project SMRITI from 09-04-2026 to 15-05-2026 and successfully submitted the proposal.

Resolved distance measurement accuracy issues in the detection system by implementing a dead-band threshold and per-frame displacement cap to eliminate YOLO centroid jitter on stationary strings. Added EMA smoothing to stabilise tracking during active movement. Retrained the detection model at 320 px inference resolution for better mark detection at range. Also added settings menu controls for toggling live feed, inference on/off, and resize width.

Adding the gmct standalone application to the current smriti fmt application.

creating a EoI document for the smriti project.
created a year wise csv with all the disease data form IDSP dataset. The year wise cdv will contain each days district wise diseases details

Implemented anti-jitter tracking so distance no longer increments while the ring is stationary (EMA smoothing + dead-zone anchor + per-frame distance cap).

User identity and permissions are now proven on every request with a server-signed token (HMAC using the server's secret key), instead of being read from a client-supplied header. The browser can no longer impersonate other users or escalate its own privileges — the server independently verifies the signature and re-checks permissions from the database

Built a complete Role Management system: roles and privileges persist in the database; create/edit/delete with live user counts.
Roles now drive routing and visible features: each role sees only what it's permitted.
Two privilege groups — Administration (user/role/subscription management, reports, system settings) and Clinical (Register, Launch FMT/GMT, Test Records, Label Data, Connect Device, etc.).