Multimodal blending for high-accuracy instance recognition

Xie, Ziang; Singh, Arjun; Uang, Justin; Narayan, Karthik; Abbeel, Pieter

doi:10.1109/iros.2013.6696666

Cited by 25 publications

(2 citation statements)

References 16 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In general, this system is detected by animal recognition and matching with binary methods for classification data. [4] Measurement of plant growth and productivity is important for the future. The survey and monitoring system used today can use UAV technology to capture photogrammetry images.…”

Section: Theoretical Basismentioning

confidence: 99%

Ecosystem Controlling Using Identification of Animal Based on Deep Learning and GLCM (Gray Level Co-occurrence Matrix) Method

Wahyu¹,

Sulianta²

2019

ujeee

View full text Add to dashboard Cite

Indonesia is the largest country in Southeast Asia. Plantations in Indonesia are still traditionally and use irrigation systems that are housed in reservoirs and ponds. While plantations in West Java generally averaged near springs and near nature preserve locations, which has a benefit impact of the available amount of water. The negative impact is entry on areas by animals and pests, which will damage Plantation grounds and other agricultural land. The animals generally go down at night when the time of the harvest season arrives. Currently, prevention of destructive animals is done by making a trap, but it is very dangerous for animals and farmers. This can lead to more severe death and destruction, especially if the dead are protected animals. One of the safe technologies, which can be used is by installing camera in the plantation area. This system is connected with the computer remotely and can operate automatically. This system is equipped with data on all types of animals that are protected. The data is processed with the Deep Learning Method and GLCM. The system will identify the animal in the category of pests or not by recognizing the pattern, texture and posture of the animal, for example if the system distinguished the wild boar. This system is considered effective and does not interfere the ecosystem and animal habitat. The addition of thermal camera and remote sensing are very effective to monitor the condition of the plant. This system is effective reduce and avoid animals killing and able to control stability the ecosystem.

show abstract

Section: Theoretical Basismentioning

confidence: 99%

Ecosystem Controlling Using Identification of Animal Based on Deep Learning and GLCM (Gray Level Co-occurrence Matrix) Method

Wahyu¹,

Sulianta²

2019

ujeee

View full text Add to dashboard Cite

show abstract

“…(7)(8)(9) However, objects in manipulation scenes are usually occluded by other objects or hands, making it impossible to fully extract global information. Xie et al (10) projected local features back onto a three-dimensional (3D) point cloud and estimated the object pose using the random sampling consensus (RANSAC) (11) algorithm. This method is robust to occlusions since global information is not required.…”

Section: Introductionmentioning

confidence: 99%

Manipulated-Object Detection and Pose Estimation Based on Multimodal Feature Points and Neighboring Patches

Tong¹,

Li²,

Zhao³

et al. 2020

Sensors and Materials

View full text Add to dashboard Cite

The detection and pose estimation of objects in human demonstrations remain challenging yet crucial tasks. The increasing availability of red-green-blue and depth sensors makes it possible to synthetize local features of color and three-dimensional (3D) geometry, which are useful for processing a wider range of objects. However, existing methods fail to combine the inherent advantages of these two features. Moreover, pose refinement methods based on whole point clouds are often affected by occlusion and background noise. In this paper, feature points of the speeded-up robust feature and the fast point feature histogram were transformed into the same 3D space. After matching them separately, multimodal feature points were jointly used to estimate a coarse pose. Subsequently, the coarse pose was refined by aligning point clouds composed of feature points' neighboring patches. During the iterative closest point process, we selected corresponding points in matched local patches. In our first and second comparative experiments, F1 scores were respectively increased by 0.1349 and 0.1633, which verified the validity of our method. Finally, the third qualitative experiment showed that the proposed method is applicable to manipulated-object detection and pose estimation.

show abstract