Javascript must be enabled to continue!
Strong Representation Learning for Weakly Supervised Object Detection
View through CrossRef
To solve the problem that the feature maps generated by feature extraction network of traditional weakly supervised learning object detection algorithm is not strong in feature, and the mapping relationship between feature space and classification results is not strong, which restricts the performance of object detection, a weakly supervised object detection algorithm based on strong representation learning is proposed in this paper. Due to enhance the representation ability of feature maps, the algorithm weighted the channels of feature maps according to the importance of each channel, to strengthen the weight of crucial feature maps and ignore the significance of secondary feature maps. Meanwhile, a Gaussian Mixture distribution model with better classification performance was used to design the object instance classifier to enhance further the representation of the mapping between feature space and classification results, and a large-margin Gaussian Mixture (L-GM) loss was designed to increase the distance between sample categories and improve the generalization of the classifier. For verifying the effectiveness and advancement of the proposed algorithm, the performance of the proposed algorithm is compared with six classical weakly supervised target detection algorithms on VOC datasets. Experiments show that the weakly supervised target detection algorithm based on strong representation learning has outperformed other classical algorithms in average accuracy (AP) and correct location (CorLoc), with increases of 1.1%~14.6% and 2.8%~19.4%, respectively.
Title: Strong Representation Learning for Weakly Supervised Object Detection
Description:
To solve the problem that the feature maps generated by feature extraction network of traditional weakly supervised learning object detection algorithm is not strong in feature, and the mapping relationship between feature space and classification results is not strong, which restricts the performance of object detection, a weakly supervised object detection algorithm based on strong representation learning is proposed in this paper.
Due to enhance the representation ability of feature maps, the algorithm weighted the channels of feature maps according to the importance of each channel, to strengthen the weight of crucial feature maps and ignore the significance of secondary feature maps.
Meanwhile, a Gaussian Mixture distribution model with better classification performance was used to design the object instance classifier to enhance further the representation of the mapping between feature space and classification results, and a large-margin Gaussian Mixture (L-GM) loss was designed to increase the distance between sample categories and improve the generalization of the classifier.
For verifying the effectiveness and advancement of the proposed algorithm, the performance of the proposed algorithm is compared with six classical weakly supervised target detection algorithms on VOC datasets.
Experiments show that the weakly supervised target detection algorithm based on strong representation learning has outperformed other classical algorithms in average accuracy (AP) and correct location (CorLoc), with increases of 1.
1%~14.
6% and 2.
8%~19.
4%, respectively.
Related Results
Analysis and Design of Work Systems using Hazard Analysis and Critical Control Points (HACCP) Standards at Mr. Sarbani’s Tempe Jombor Sukoharjo
Analysis and Design of Work Systems using Hazard Analysis and Critical Control Points (HACCP) Standards at Mr. Sarbani’s Tempe Jombor Sukoharjo
<p><strong>MSMEs </strong><strong>have very </strong><strong>large</strong><strong> number of actors so </strong><strong>...
Representasi Gender dalam Folklor Jepang
Representasi Gender dalam Folklor Jepang
<p><em>Abstrak</em><strong> - </strong><strong>Penelitian ini bertujuan untuk menemukan representasi gender dalam folklor Jepang. Data utama dal...
Makna Puisi Kotoba (言葉) Karya Tanikawa Shuntaro: Analisis Semiotika Riffa Terre
Makna Puisi Kotoba (言葉) Karya Tanikawa Shuntaro: Analisis Semiotika Riffa Terre
<p><em>Abstrak</em> - <strong>Penelitian ini bertujuan untuk untuk menemukan makna dalam puisi <em>Kotoba</em> karya Tanikawa Shuntaro.</stro...
Motivasi Berjilbab Mahasiswi Universitas Al Azhar Indonesia (UAI)
Motivasi Berjilbab Mahasiswi Universitas Al Azhar Indonesia (UAI)
<p><em>Abstrak – </em><strong>Titik tolak penelitian ini adalah untuk menelusuri secara ilmiah motivasi berjilbab mahasiswi UAI –khususnya mereka yang berji...
Penyusunan Model Korpus Al-Qur’an Digital
Penyusunan Model Korpus Al-Qur’an Digital
<p><em>Abstrak – </em><strong>Penelitian ini bertujuan menyusun sebuah model file korpus Al-Qur'an digital yang dapat digunakan sebagai bahan data primer ba...
Rancangan Service Scorecard sebagai Pengukuran Kinerja Pelayanan Cargo
Rancangan Service Scorecard sebagai Pengukuran Kinerja Pelayanan Cargo
<p><em>Abstrak<strong> – </strong></em><strong>Persaingan antara pelaku bisnis </strong><strong><em>charter flight</em>&...
Potensi Daun Teh (Camellia sinensis) dan Daun Anting-anting Acalypha indica L. dalam Menghambat Pertumbuhan Salmonella typhi
Potensi Daun Teh (Camellia sinensis) dan Daun Anting-anting Acalypha indica L. dalam Menghambat Pertumbuhan Salmonella typhi
<em><em></em></em><em>Abstrak <strong>-</strong> <strong>Salmonella typhi </strong></em><strong>adalah bakteri pen...
Tipologi Keislaman Jamaah Masjid Agung Al-Azhar (MAA)
Tipologi Keislaman Jamaah Masjid Agung Al-Azhar (MAA)
<p><em>Abstrak – </em><strong>Titik tolak penelitian ini adalah untuk mendeskripsikan secara ilmiah tipologi keislaman jamaah Masjid Agung Al-Azhar (MAA), J...

