Related Experiment Videos
LeanCOD: Real-Time Small Camouflaged Object Detection on Edge Devices
Youngjin Kim1, Dong He1, Young Hoo Cho1
1Research and Development Center, dSPECTER, Seongnam 13449, Republic of Korea.
Sensors (Basel, Switzerland)
|July 28, 2026
Summary
LeanCOD improves camouflaged object detection (COD) for small objects and enables real-time performance on edge devices. This framework achieves high accuracy and speed, outperforming existing methods.
Area of Science:
- Computer Vision
- Artificial Intelligence
Background:
- Camouflaged object detection (COD) methods struggle with small objects and real-time edge device inference.
- Existing methods show significant performance degradation on extra-small objects (0-1% size regime).
Purpose of the Study:
- To develop a high-resolution camouflaged object detection framework capable of real-time inference on edge devices.
- To specifically address the performance bottleneck caused by small camouflaged objects.
Main Methods:
- Proposed LeanCOD framework pairing a strong foundation-model encoder with a lightweight decoder.
- Implemented a size-aware composite loss function to enhance supervision on small objects.
- Evaluated performance on the COD10K dataset and deployed on an NVIDIA Jetson AGX Orin using TensorRT FP16.
Main Results:
- LeanCOD achieved an Sα score of 0.915 on COD10K at 576×576 resolution, surpassing existing methods.
- Achieved real-time inference at 31.6 FPS on an NVIDIA Jetson AGX Orin with an Sα of 0.908 at 576×576 resolution.
- Demonstrated superior performance, particularly in the challenging extra-small object regime.
Conclusions:
- LeanCOD effectively addresses limitations in camouflaged object detection for small objects.
- The framework enables efficient, high-resolution, real-time COD on edge computing platforms.
- LeanCOD represents a significant advancement for practical applications of camouflaged object detection.
Related Concept Videos
01:05Difference from Background: Limit of Detection
The limit of detection (LOD) is the smallest amount of analyte that can be distinguished from the background noise. The LOD value corresponds to the concentration at which the analyte signal is three times larger than the standard deviation of the blank signal. Below this value, the analyte signal cannot be differentiated from the background noise. It is calculated by dividing the calibration slope by 3 times the standard deviation of the blank signals.
The LOD indicates the presence or absence...
The LOD indicates the presence or absence...
02:16Light Acquisition
In order to produce glucose, plants need to capture sufficient light energy. Many modern plants have evolved leaves specialized for light acquisition. Leaves can be only millimeters in width or tens of meters wide, depending on the environment. Due to competition for sunlight, evolution has driven the evolution of increasingly larger leaves and taller plants, to avoid shading by their neighbors with contaminant elaboration of root architecture and mechanisms to transport water and nutrients.
01:10Detection of Black Holes
Although black holes were theoretically postulated in the 1920s, they remained outside the domain of observational astronomy until the 1970s.
Their closest cousins are neutron stars, which are composed almost entirely of neutrons packed against each other, making them extremely dense. A neutron star has the same mass as the Sun but its diameter is only a few kilometers. Therefore, the escape velocity from their surface is close to the speed of light.
Not until the 1960s, when the first neutron...
Their closest cousins are neutron stars, which are composed almost entirely of neutrons packed against each other, making them extremely dense. A neutron star has the same mass as the Sun but its diameter is only a few kilometers. Therefore, the escape velocity from their surface is close to the speed of light.
Not until the 1960s, when the first neutron...
01:30Electronic Distance Measuring Instruments
Electronic Distance Measuring Instruments (EDMs) are essential tools in modern surveying, offering precise distance measurements by emitting electromagnetic signals and calculating the time required for these signals to travel to a target and return. Two primary types of signals are used in EDMs — light waves and microwaves — each suited to specific environmental and distance requirements. Light-wave-based EDMs utilize either infrared or laser light, providing high accuracy over short distances...