Compression

• ICME
• 2023

While conventional image compression techniques are optimized for human visual perception, the rise of machine learning techniques has led to the emergence of image compression methods tailored for machine vision tasks. Although a few recent studies explored target-dependent reconfiguration of lightweight codecs such as JPEG, these approaches are limited to specific trained bitrates only. Moreover, existing deep learning-based compression frameworks entail a high computational cost, making them impractical for real-time compression on devices with limited resources. In this paper, we present a novel JPEG compression framework that can adaptively generate an optimal quantization table (QT) depending on both the target bitrate and the target metric (quality or accuracy). To provide fine controllability over a wide range of bitrates, we employ a feature modulation technique to a QT generator and bitrate predictor, which are trained by a novel training method called bitrate range partitioning. Our simulation results show that the proposed framework enhances the performance of standard JPEG by up to 2dB in PSNR and 10% in accuracy at the same bitrate, while incurring minimal computational overhead compared to JPEG.

• ICME
• 2023

A sensor network is an underpinning infrastructure that enables various future IoT applications, such as precision agriculture, smart farm, and greenhouse monitoring. However, these sensor devices often suffer from short-lived battery lifetime that incurs frequent maintenance operation. Although there have been a few attempts to smartly reduce the power consumption associated with communication tasks of the sensors, very few have addressed the power consumption of sensing tasks. In light of this shortcoming, we propose an information-aware sensing framework that adaptively adjusts the sensing interval for energy-saving operations based on the learned behavior of the sensor data. To prove the effectiveness of the proposed framework, we have deployed four BLE beacons equipped with luminosity and temperature sensors to collect real-life data from a desert greenhouse, which is then used to train and evaluate our proposed framework. Additionally, we have implemented the proposed framework on a commodity BLE beacon device to validate the energy-saving performance of the proposed framework. The results demonstrate that the proposed framework can effectively reduce the energy consumption involved in sensing tasks by 30% and extend the battery lifetime by up to 75%.