This document presents CNNECST, an automated framework for hardware acceleration of convolutional neural networks (CNNs) on FPGAs. The framework bridges machine learning frameworks and FPGA design through high-level APIs, an intermediate representation, and C++ libraries. It was tested on two datasets and FPGAs, achieving speedups of 45x over CPUs and higher energy efficiency, while maintaining accuracy. Challenges include supporting more layer types and reduced precision data.