Learning to detect and localize many objects from few examples

Bastien Moysset 1, 2 Christopher Kermorvant 3 Christian Wolf 2
2 imagine - Extraction de Caractéristiques et Identification
LIRIS - Laboratoire d'InfoRmatique en Image et Systèmes d'information
Abstract : The current trend in object detection and localization is to learn predictions with high capacity deep neural networks trained on a very large amount of annotated data and using a high amount of processing power. In this work, we propose a new neural model which directly predicts bounding box coordinates. The particularity of our contribution lies in the local computations of predictions with a new form of local parameter sharing which keeps the overall amount of trainable parameters low. Key components of the model are spatial 2D-LSTM recurrent layers which convey contextual information between the regions of the image. We show that this model is more powerful than the state of the art in applications where training data is not as abundant as in the classical configuration of natural images and Imagenet/Pascal VOC tasks. We particularly target the detection of text in document images, but our method is not limited to this setting. The proposed model also facilitates the detection of many objects in a single image and can deal with inputs of variable sizes without resizing.
Type de document :
Rapport
[Research Report] INSA-Lyon; LIRIS; A2IA. 2016
Liste complète des métadonnées

https://hal.archives-ouvertes.fr/hal-01399159
Contributeur : Christian Wolf <>
Soumis le : vendredi 18 novembre 2016 - 13:41:42
Dernière modification le : samedi 19 novembre 2016 - 01:10:06

Identifiants

  • HAL Id : hal-01399159, version 1
  • ARXIV : 1611.05664

Collections

Citation

Bastien Moysset, Christopher Kermorvant, Christian Wolf. Learning to detect and localize many objects from few examples. [Research Report] INSA-Lyon; LIRIS; A2IA. 2016. <hal-01399159>

Partager

Métriques

Consultations de la notice

84