Enquire Now
Home/ AI / Embedded/ TinyML Projects
50+ TinyML Project Topics · Keyword Spotting · Vision · Sensors · Optimization · Bangalore 2026

TinyML Projects

Audio · Vision · Sensors · Anomaly Detection · Model Optimization · Hardware Deployment — Best final-year and research topics for machine learning on microcontrollers. TensorFlow Lite Micro, Edge Impulse, Arduino, STM32 and ESP32. Firmware, metrics, report, PPT and viva support from Bangalore.

Browse 50+ Topics Talk to an Expert
52+
TinyML Topics
8
Application Domains
9800+
Students Guided
Audio / KWS Vision Sensors / IMU Anomaly Detection Optimization Hardware Platforms Applications Advanced

TinyML Final Year Projects 2026

TinyML runs machine learning models on resource-constrained microcontrollers (KB of RAM, mW power). Projects typically train a small model, convert it to TensorFlow Lite, deploy with TFLite Micro or Edge Impulse, and report accuracy, latency, flash and RAM usage.

Platforms: Arduino Nano 33 BLE Sense, ESP32, STM32, Raspberry Pi Pico; tools: TensorFlow/Keras, TFLite Micro, Edge Impulse, CMSIS-NN.

Edge AI TinyML Projects and Course

Tools & Platforms Used
TFLite Micro Edge Impulse Arduino STM32Cube.AI ESP32 TensorFlow / Keras

Best TinyML Project Topics 2026

52 topics across major TinyML application domains with tools used.

#TinyML Project TopicTools / Platforms
🎤 Audio · Keyword Spotting · Sound Classification
01AudioKeyword Spotting (Yes/No / Custom Wake Word) on MicrocontrollerTFLite Micro · Arduino · Google Speech Commands
02AudioEnvironmental Sound Classification (UrbanSound / ESC-50 subset)Edge Impulse · TFLite Micro · ESP32
03AudioCough / Sneeze / Snore Detection for Health Monitoring WearableEdge Impulse · Arduino Nano 33 · mic
04AudioGlass-Break / Alarm Sound Detection for Security Edge NodeTFLite Micro · ESP32 · MFCC features
05AudioBaby Cry Detection and Alert System on Low-Power MCUEdge Impulse · Arduino · BLE notify
06AudioMulti-Keyword Spotting with DS-CNN / MicroNet ArchitectureTFLite Micro · Keras · CMSIS-NN
07AudioVoice Activity Detection (VAD) Pipeline Before KWS StageTFLite Micro · Arduino · energy + ML
📷 Vision · Image Classification · Object Detection
08VisionPerson / No-Person Detection with Tiny Vision Model on MCU CameraEdge Impulse · OV7670 / HM01B0 · Arduino
09VisionHand Gesture Classification from Low-Resolution Camera FramesTFLite Micro · ESP32-CAM · CNN
10VisionDigit / Character Recognition on Tiny Display CaptureEdge Impulse · MNIST-style · MCU
11VisionPlant Disease Leaf Classification on Edge DeviceTFLite Micro · custom dataset · ESP32-CAM
12VisionFace Mask Detection with Quantized MobileNet-style ModelTFLite Micro · Edge Impulse · camera
13VisionObject Counting (e.g. Products on Conveyor) with Tiny DetectorEdge Impulse FOMO · Arduino / STM32
14VisionVisual Wake Words Benchmark Replication on Constrained HardwareTFLite Micro · VWW dataset · metrics
📐 Sensors · IMU · Gesture · Motion
15SensorIMU Gesture Recognition (Swipe, Circle, Shake) on WearableEdge Impulse · Arduino Nano 33 · IMU
16SensorHuman Activity Recognition (Walk, Run, Sit) from AccelerometerTFLite Micro · HAR datasets · MCU
17SensorFall Detection System for Elderly Care using IMU + TinyMLEdge Impulse · BLE · Arduino
18SensorRepetition Counting for Exercise / Gym Form FeedbackIMU · TFLite Micro · ESP32
19SensorMouse / Air-Mouse Control via Hand Motion ClassificationIMU · Edge Impulse · HID concepts
20SensorVibration Pattern Classification for Machine State MonitoringAccelerometer · TFLite Micro · STM32
21SensorMulti-Sensor Fusion: IMU + Audio for Context AwarenessEdge Impulse · Nano 33 BLE Sense
⚠️ Anomaly Detection · Predictive Maintenance
22AnomalyMotor / Fan Anomaly Detection from Current or Vibration SensorsEdge Impulse · autoencoder / GMM · MCU
23AnomalyTemperature / Humidity Drift Anomaly Alert on IoT NodeTFLite Micro · ESP32 · time-series
24AnomalyBearing Fault Detection with Tiny Spectral Features + ClassifierEdge Impulse · FFT · Arduino
25AnomalyPower Consumption Anomaly Detection for Smart Plug MCUCurrent sensor · TFLite Micro · ESP32
26AnomalyAcoustic Anomaly Detection in Industrial EquipmentMic · Edge Impulse · unsupervised
27AnomalyComparative Study: Supervised vs Unsupervised Tiny Anomaly ModelsEdge Impulse · TFLite Micro · metrics
⚙️ Model Optimization · Quantization · Compression
28OptPost-Training Quantization (INT8) Impact on Accuracy and SizeTensorFlow · TFLite · MCU deploy
29OptQuantization-Aware Training for MCU Keyword Spotting ModelKeras · QAT · TFLite Micro
30OptPruning + Quantization Pipeline for Flash/RAM BudgetTensorFlow Model Optimization · MCU
31OptLatency vs Accuracy Trade-off Study across Model SizesTFLite Micro · cycle counters · report
32OptCMSIS-NN Optimised Kernels vs Generic TFLite Micro RuntimeCMSIS-NN · STM32 · benchmarks
33OptKnowledge Distillation from Large Teacher to Tiny Student ModelKeras · TFLite · MCU validation
34OptMemory Footprint Analysis: Arena Size, Tensor Allocation, Peak RAMTFLite Micro · profiling · Arduino
🔧 Hardware Platforms · Firmware · Deployment
35HWEnd-to-End Deploy: Train → TFLite → Arduino Nano 33 BLE SenseArduino · TFLite Micro · Edge Impulse
36HWESP32 TinyML Pipeline with Wi-Fi Result ReportingESP-IDF / Arduino · TFLite · MQTT
37HWSTM32Cube.AI Conversion and Deployment on Nucleo BoardSTM32Cube.AI · X-CUBE-AI · Nucleo
38HWRaspberry Pi Pico (RP2040) TinyML Inference with Micropython or CPico · TFLite Micro · C SDK
39HWDual-Core / DMA Optimisation for Continuous Audio InferenceESP32 · dual core · ring buffer
40HWPower Measurement of Inference vs Idle for Battery Lifetime EstimateCurrent probe · MCU · report
🏠 Domain Applications · Smart Home · Health · Industry
41AppSmart Home Keyword Control (Lights / Fan) with On-Device KWSArduino · relays · TFLite Micro
42AppOccupancy Detection for Room Energy Saving using PIR + Tiny VisionEdge Impulse · ESP32-CAM · control
43AppWildlife / Bird Call Classification on Solar-Powered Edge NodeEdge Impulse · ESP32 · low power
44AppHand Hygiene Compliance Monitoring with Gesture TinyMLIMU · Edge Impulse · wearable
45AppSmart Agriculture: Soil / Leaf Condition Classification on Field MCUSensors · TFLite Micro · LoRa concepts
46AppFactory Safety: PPE Detection (Helmet) with Tiny Vision ModelEdge Impulse FOMO · camera · alert
47AppParking Slot Occupancy Sensing with Ultrasonic + Tiny ClassifierUltrasonic · MCU · TFLite
🔬 Advanced · Research · Systems
48AdvContinual / Incremental Learning on MCU with Replay Buffer LimitsTFLite Micro · custom training loop
49AdvFederated Learning Concepts with Multiple Edge Nodes (Simulated)Python sim · TFLite · aggregation
50AdvSecure Inference: Model Encryption / Secure Boot ConsiderationsMCU secure boot · TFLite · notes
51AdvBenchmark Suite: Same Model across Arduino, ESP32 and STM32Multi-board · latency · RAM · accuracy
52AdvCapstone: Full TinyML Product — Dataset, Train, Optimise, Deploy, DashboardEdge Impulse / TFLite · MCU · report

Topics follow common university TinyML curricula and Edge Impulse / TFLite Micro practice. Contact us for training notebooks, firmware sketches, hardware notes, university-format report, PPT and viva Q&A.

Why Choose Us for TinyML Projects?

Bangalore-based guidance for BE, BTech and MTech students deploying ML on microcontrollers.

Audio & Keyword Spotting

Wake-word and sound classification pipelines with MFCC features and on-device latency reporting.

Vision on MCU

Person detection, gestures and FOMO-style object counting with camera modules and quantised CNNs.

Sensors & Anomaly

IMU activity recognition, fall detection and predictive-maintenance anomaly models on the edge.

Optimization & Deploy

INT8 quantisation, pruning, CMSIS-NN and multi-board benchmarks with clear memory and power metrics.

Frequently Asked Questions — TinyML Projects

Strong topics include keyword spotting, person detection with tiny vision models, IMU gesture/activity recognition, vibration anomaly detection, INT8 quantisation studies, and full deploy on Arduino Nano 33, ESP32 or STM32 with latency and RAM reporting.
TensorFlow Lite Micro, Edge Impulse, Arduino IDE/PlatformIO, STM32Cube.AI, ESP-IDF; boards: Arduino Nano 33 BLE Sense, ESP32/ESP32-CAM, STM32 Nucleo, Raspberry Pi Pico.
Yes. Packages include training notebooks, TFLite conversion steps, firmware sketches, hardware wiring notes, university-format report, PPT and viva Q&A on memory, latency and accuracy trade-offs.
TinyML targets microcontrollers with KB of RAM and no OS (or bare-metal RTOS), using TFLite Micro. Raspberry Pi / Jetson run Linux and larger TFLite or TensorRT models — still edge, but not “tiny”.