C3(Crowd-Counting-Code): An open source Crowd- Counting framework in PyTorch
Some key takeaways from the research on the C3 framework
This figure demonstrates the data flow and architecture of C3F. It can be divided into two parts, i.e., data preprocessing strategy and model baseline.
Goal
The purpose of C3 is an efficient, flexible framework for supervised crowd counting. At the same time, it provides the performances of some basic networks and classic algorithms on the mainstream datasets.
Features
- Convenient development kit: It is a convenient dev kit on the six maintream datasets UCF CC 50, WorldExpo’10, SHT A(ShanghaiTech-A), SHT B(ShanghaiTech-B), UCF-QNRF, and GCC.
- Solid baselines: It provides some baselines of some classic pre-trained models, such as AlexNet, VGG, ResNet and so on. Base on it, you can easily compare your proposed models’ effects with them.
- Powerful log: It does not only record the loss, visualization in Tensorboard, but also save the current code package (including parameters settings). The saved code package can be directly ran to reproduce the experiments at any time. You won’t be bothered by forgetting the confused parameters.
Input Size Specifications
Label Transformation: C3F provides two operations for label transformation, including ground truth scale down-sampling and label normalization.
Results:
To implement C3 on your system, you will require CUDA. Make sure that PyTorch is compiled for CUDA.
Conclusion: The C3F is designed to reduce human cost in training process and offers good results.
Link to research paper: https://arxiv.org/pdf/1907.02724v1.pdf
Link to GitHub code: https://github.com/gjy3035/C-3-Framework
