Dobiasd/frugally-deep
A lightweight header-only library for using Keras (TensorFlow) models in C++.
This tool helps C++ developers integrate Keras (TensorFlow) machine learning models directly into their C++ applications. You can build and train your model in Python using Keras, then convert it to a lightweight format. The C++ application then takes this converted model and uses it to make predictions, without needing to link against the full TensorFlow library. This is ideal for C++ developers building high-performance or resource-constrained applications that need to leverage pre-trained AI models.
1,119 stars.
Use this if you need to run machine learning model predictions within a C++ application and want to avoid the overhead of the full TensorFlow library.
Not ideal if you need to perform model training or utilize GPUs for prediction within your C++ application, or if your Keras model uses unsupported layer types like recurrent neural networks (RNNs) or custom Lambda layers.
Stars
1,119
Forks
236
Language
C++
License
MIT
Category
Last pushed
Dec 04, 2025
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/ml-frameworks/Dobiasd/frugally-deep"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Compare
Related frameworks
flashlight/flashlight
A C++ standalone library for machine learning
wichtounet/dll
Fast Deep Learning Library (DLL) for C++ (ANNs, CNNs, RBMs, DBNs...)
NVlabs/tiny-cuda-nn
Lightning fast C++/CUDA neural network framework
KasperskyLab/knp
Kaspersky Neuromorphic Platform
tiny-dnn/tiny-dnn
header only, dependency-free deep learning framework in C++14