v1.1.7

Fixed PyPi package bug

Had incidentally left scikit-python in and omitted matplotlib from the PyPi package.
Small fix in VPRTempo data loader to remove the shuffle argument.

v1.1.6

Updates in v1.1.6

• Modified information for the Oxford RobotCar dataset
• Removed activation layers inbetween linear feature and output layers which improves performance
• Allow users to skip the first n number of images in a dataset easier with the --skip argument
• Split .csv files for image loading into unique dataset directories, better flexibility for datasets without matching image names

What's Changed

• nn.Sequential fix by @AdamDHines in #13
• Lfs update & removal by @AdamDHines in #16

Full Changelog: v1.1.5...v1.1.6

v1.1.5

Bugs fixed in v1.1.5

• Issue with large models and CUDA, previous version loaded all models at once using CUDA memory very quickly. FIX: Models are initialized and stored on CPU after creation, moved back to CUDA when training/inferencing. This will have impact on inferencing performance since models are being moved back and forward from CPU<->GPU but this only affects very large models.

Features added in v1.1.5

• Easier installation with conda-forge packaging, can now simply install with conda install vprtempo -c conda-forge
• Added options to plot metrics, similarity matrices, and additionally output precision and recall in json file
• Modified model names so that they're clearer, also rather than model names simply being network architecture dims more information such as the datasets themselves included so multiple models with the same architecture can be trained across different datsets without overwriting them.

v1.1.4

Modify __init__ imports, change version number

v1.1.3

• Modified codebase into a singular module
• Easing module import (e.g. for blitnet, metrics, etc)
• Needed for conda-forge deployment

v1.1.2

Fixing repo structure for conda-forge release

v1.1.1

• Minor fixes to PyPi packaging

v1.1.0

• Full integration of VPRTempo into torch.nn architecture
• Quantization Aware Training (QAT) enabled to train weights in int8 space
• Addition of tutorials in Jupyter Notebooks to learn how to use VPRTempo as well as explain the computational logic
• Simplification of weight operations, reducing to a single weight tensor - allowing positive and negative connections to change sign during training
• Easier dependency installation with PyPi/pip
• And more!

v1.0.0

VPRTempo code for ICRA 2024 submission.

What's Changed

• V1.0.0 alpha to v1.0.0 full main merge by @AdamDHines in #5

Full Changelog: v0.0.1...v1.0.0

v0.0.1

VPRTempo v0.0.1: non-modular temporally encoded spiking neural network for visual place recognition