本文介绍了一种结合行人检测的卷积网络模型，通过无监督的卷积稀疏自编码器预训练不同层级特征，并采用端到端的监督训练方式来调整特征和训练分类器。此外，该模型还利用多阶段特征与层间跳接连接，实现全局形状检测器与局部图案检测器的有效融合。
Pedestrian detection + a convolutional network model （with a few twists）.
1、Unsupervised convolutional sparse auto-encoders to pre-train features at all levels from the relatively small INNRIA dataset.
2、End-to-end supervised training to train the classifier and fine-tune the features in an intregrated fashion.
3、Multi-stage features with layer-skipping connections enable output stages to combine global shape detectors with local motif detectors.
扫一扫
593
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
