20211027

Updated with support for TF 2.6.0, OpenCV 3.4.16 and 4.5.4. Max CUDA updated to 11.4.2

Full list of available containers (up to this release) can be seen at:
https://github.com/datamachines/cuda_tensorflow_opencv/blob/master/Builds-DockerHub.md

For list of installed pip packages, for example PyTorch, see the matching entry in the BuildInfo-TensorFlow directory.

20210812

Updated with support for TF 2.6.0, OpenCV 3.4.15 and 4.5.3. Max CUDA updated to 11.4.1

Full list of available containers (up to this release) can be seen at:
https://github.com/datamachines/cuda_tensorflow_opencv/blob/master/Builds-DockerHub.md

20210810

Updated with support for TF 2.5.1, OpenCV 3.4.15 and 4.5.3

Full list of available containers (up to this release) can be seen at:
https://github.com/datamachines/cuda_tensorflow_opencv/blob/master/Builds-DockerHub.md

20210601

• Include TF 2.5.0 builds for CPU and CUDA10,11
• Providing both CUDA 11.2 and 11.3 builds (to use 11.3 you will need the latest Nvidia drivers)

Full list of available containers (up to this release) can be seen at:
https://github.com/datamachines/cuda_tensorflow_opencv/blob/master/Builds-DockerHub.md

20210420

CPU + CUDA 9.2, 10.2, 11.2.2
OpenCV: 3.4.14, 4.5.2
TensorFlow: 1.15.5, 2.4.1

Full list of available containers (up to this release) can be seen at:
https://github.com/datamachines/cuda_tensorflow_opencv/blob/master/Builds-DockerHub.md

20210414

CUDA: 9.2 (Ubuntu 18.04), 10.2 (Ubuntu 18.04), 11.2.0 (Ubuntu 20.04)
OpenCV: 3.4.14, 4.5.2
TensorFlow: 1.15.5, 2.4.1

As of 20210414, we are obsoleting the cuda_tensorflow_opencv container in favor of only tensorflow_opencv and cudnn_tensorflow_opencv

Full list of available containers (up to this release) can be seen at:
https://github.com/datamachines/cuda_tensorflow_opencv/blob/master/Builds-DockerHub.md

20210211

CUDA: 9.2 (Ubuntu 18.04), 10.2 (Ubuntu 18.04), 11.2.0 (Ubuntu 20.04)
OpenCV: 3.4.13, 4.5.1
TensorFlow: 1.15.5, 2.4.1

Available containers from this release:
https://hub.docker.com/r/datamachines/tensorflow_opencv
datamachines/tensorflow_opencv:2.4.1_4.5.1-20210211
datamachines/tensorflow_opencv:2.4.1_3.4.13-20210211
datamachines/tensorflow_opencv:1.15.5_4.5.1-20210211
datamachines/tensorflow_opencv:1.15.5_3.4.13-20210211

https://hub.docker.com/r/datamachines/cuda_tensorflow_opencv
datamachines/cuda_tensorflow_opencv:11.2.0_2.4.1_4.5.1-20210211
datamachines/cuda_tensorflow_opencv:11.2.0_2.4.1_3.4.13-20210211
datamachines/cuda_tensorflow_opencv:10.2_2.4.1_4.5.1-20210211
datamachines/cuda_tensorflow_opencv:10.2_2.4.1_3.4.13-20210211
datamachines/cuda_tensorflow_opencv:10.2_1.15.5_4.5.1-20210211
datamachines/cuda_tensorflow_opencv:10.2_1.15.5_3.4.13-20210211
datamachines/cuda_tensorflow_opencv:9.2_2.4.1_4.5.1-20210211
datamachines/cuda_tensorflow_opencv:9.2_2.4.1_3.4.13-20210211
datamachines/cuda_tensorflow_opencv:9.2_1.15.5_4.5.1-20210211
datamachines/cuda_tensorflow_opencv:9.2_1.15.5_3.4.13-20210211

https://hub.docker.com/r/datamachines/cudnn_tensorflow_opencv
datamachines/cudnn_tensorflow_opencv:11.2.0_2.4.1_4.5.1-20210211
datamachines/cudnn_tensorflow_opencv:11.2.0_2.4.1_3.4.13-20210211
datamachines/cudnn_tensorflow_opencv:10.2_2.4.1_4.5.1-20210211
datamachines/cudnn_tensorflow_opencv:10.2_2.4.1_3.4.13-20210211
datamachines/cudnn_tensorflow_opencv:10.2_1.15.5_4.5.1-20210211
datamachines/cudnn_tensorflow_opencv:10.2_1.15.5_3.4.13-20210211
datamachines/cudnn_tensorflow_opencv:9.2_1.15.5_4.5.1-20210211
datamachines/cudnn_tensorflow_opencv:9.2_1.15.5_3.4.13-20210211

Full list of available containers (up to this release) can be seen at:
https://github.com/datamachines/cuda_tensorflow_opencv/blob/master/Builds-DockerHub.md

20201204

Supports:
CUDA: 9.2 or 10.2
OpenCV: 3.4.12 or 4.5.0
TensorFlow: 1.15.4 or 2.3.1

Note: added support for Python 3.7 for TensorFlow 1 builds and Python 3.8 for Tensorflow 2 builds (makes use of the deadsnakes/ppa and changes the default python3). Warning: only pip3 installed packages will work, not apt-get installed ones (Python 3.6 is still the default for Ubuntu 18.04)

20200803

Added PyTorch

Supports:
CUDA: 9.2 or 10.2
OpenCV: 3.4.11 or 4.4.0
TensorFlow: 1.15.3 or 2.3.0

Note: Removal of cudnn_ version for CUDA 9.2 with TensorFlow 2.3.0
Note: TensorFlow will not have GPU support unless it was compiled with CUDNN support.

20200615

TensorFlow is now built from source.

CUDA: 9.2 or 10.2
OpenCV: 3.4.10 or 4.3.0
TensorFlow: 1.15.3 or 2.2.0

Note that TensorFlow will not have GPU support unless it was compiled with CUDNN support.