Learned-Lensless-3D-Camera

Open source code and demonstrations of learned lensless 3D camera

Paper

Clone this repository:

git clone https://github.com/Yang-Research-Laboratory/Learned-Lensless-3D-Camera.git

Pre-trained models

We provide multiple pre-trained reconstruction modules (reconM) and enhancement module (enhanceM) for 3D imaging demonstration. The pre-trained models can be found in shared models folder, the models reconM contains 14 physics-aware models trained at object distances from 10-60cm. The enhanceM is pre-trained from shared database.

Test dataset

We provide several raw measurements of real objects by our 3D camera as test data for demonstrating 3D imaging in shared folder.
To test 3D imaging demonstration, download the test datasets into the datasets folder at same directory, and run the code test_3Dimaging.py.
To test see-through targets demonstration, download the test datasets into the datasets folder at same directory, and run the code test_seethrough.py.

Training your own models

To train the models on your customized imagers, you can generated models from our templates, specify the size and format of input images in our example code.