Support for semantic segmentation and SSD networks with samples.Many new layers, including deconvolution, LSTM etc.Greatly improved and accelerated dnn module in opencv_contrib:.There have been many contributions besides GSoC: Anguelos Nicolaou (Lluis Gomez) – Word-spotting CNN based algorithm.Yida Wang (Manuele Tamburrano, Stefano Fabri) – Quantization and semantic saliency detection with tiny_dnn.Edgar Riba (Manuele Tamburrano, Stefano Fabri) – tiny_dnn improvements and integration.Use names like “myfilestorage.xml?base64” when writing file storage to store big chunks of numerical data in base64-encoded form.
Merry Christmas and Happy New Year! Changes 969 patches have been merged and 478 issues (bugs & feature requests) have been closed.īig thanks to everyone who participated! If you contributed something but your name is missing, please, let us know. 1 year after 3.1 release and after the OpenCV core team has moved back to Intel we are pleased to announce OpenCV 3.2 release, with tons of improvements and bug fixes.