📄️ TensorFlow Lite
Ship a plant disease classifier that runs offline on farmers' phones. Conversion, quantization, delegates, Android and iOS, latency budgets. Duration: 4h, 40-question exam and verifiable certificate.
📄️ 1. On-device constraints
Module 1 of the TensorFlow Lite premium course: memory, compute, energy, offline operation, perceived latency and privacy — the budget that governs every later choice.
📄️ 2. Converting to TFLite
Module 2 of the TensorFlow Lite premium course: TFLiteConverter from SavedModel and Keras, supported operators, Select TF Ops, signatures and metadata.
📄️ 3. Post-training quantization
Module 3 of the TensorFlow Lite premium course: three post-training quantization modes, the representative dataset, and their measured effect on size, latency and accuracy.
📄️ 4. Quantization-aware training
Module 4 of the TensorFlow Lite premium course: simulate int8 during training, recover the accuracy lost to post-training quantization, and know when the extra cost is worth it.
📄️ 5. Pruning and size reduction
Module 5 of the TensorFlow Lite premium course: magnitude pruning, sparsity schedules, weight clustering, and the compression that turns sparse weights into a smaller file.
📄️ 6. Interpreter and delegates
Module 6 of the TensorFlow Lite premium course: interpreter lifecycle, input and output tensors, GPU, NNAPI, Core ML and XNNPACK delegates, and how the CPU fallback works.
📄️ 7. Android integration
Module 7 of the TensorFlow Lite premium course: Gradle dependencies, model loading, camera preprocessing, background inference and the Task Library shortcut.
📄️ 8. iOS integration
Module 8 of the TensorFlow Lite premium course: CocoaPods and Swift Package integration, model loading, Core ML in comparison, image preprocessing and camera permissions.
📄️ 9. Measuring latency and energy
Module 9 of the TensorFlow Lite premium course: the benchmark tool, warm-up, percentiles instead of averages, energy per inference and results across three phone tiers.
📄️ 10. On-device project
Module 10 of the TensorFlow Lite premium course: assemble a complete leaf-disease classifier, defend the variant chosen, plan updates and document the trade-offs sacrificed.
📄️ Recap and exam
Complete recap of the TensorFlow Lite premium course: constraints, conversion, quantization, pruning, delegates, Android and iOS, measurement, then the 40-question exam.