High Speed Object Detection

Supervisor: Talha Uddin Sheikh

Study and implementation of different pruning and quantization techniques for deep learning model compression for state of the art object detection models.

Tasks

• Study of different pruning methods for deep learning models
• Study of different quantization methods for deep learning models
• Implementation of different pruning and quantization methods on a state of the art object detection model
• Experimentation and benchmarking

Related Literature

[1] Helms, D., Amende, K., Bukhari, S., de Graaff, T., Frickenstein, A., Hafner, F., … & Vemparala, M. R. (2021, July). Optimizing Neural Networks for Embedded Hardware. In SMACD/PRIME 2021; International Conference on SMACD and 16th Conference on PRIME (pp. 1-6). VDE.
[2] Hawks, B., Duarte, J., Fraser, N. J., Pappalardo, A., Tran, N., & Umuroglu, Y. (2021). Ps and qs: Quantization-aware pruning for efficient low latency neural network inference. Frontiers in Artificial Intelligence, 4, 676564.
[3] Hubara, I., Courbariaux, M., Soudry, D., El-Yaniv, R., & Bengio, Y. (2017). Quantized neural networks: Training neural networks with low precision weights and activations. The Journal of Machine Learning Research, 18(1), 6869-6898.