Forrest N. Iandola, Anting Shen, Peter Gao, Kurt Keutzer
Recently, there has been a flurry of industrial activity around logo recognition, such as Ditto's service for marketers to track their brands in user-generated images, and LogoGrab's mobile app platform for logo recognition. However, relatively little academic or open-source logo recognition progress has been made in the last four years. Meanwhile, deep convolutional neural networks (DCNNs) have revolutionized a broad range of object recognition applications. In this work, we apply DCNNs to logo recognition. We propose several DCNN architectures, with which we surpass published state-of-art accuracy on a popular logo recognition dataset.
| Task | Dataset | Metric | Value | Model |
|---|---|---|---|---|
| Object Detection | FlickrLogos-32 | MAP | 74.4 | DeepLogo (VGG) |
| Object Detection | FlickrLogos-32 | MAP | 73.5 | DeepLogo (AlexNet) |
| Image Classification | FlickrLogos-32 | Accuracy | 89.6 | DeepLogo (GoogLeNet-GP) |
| 3D | FlickrLogos-32 | MAP | 74.4 | DeepLogo (VGG) |
| 3D | FlickrLogos-32 | MAP | 73.5 | DeepLogo (AlexNet) |
| 2D Classification | FlickrLogos-32 | MAP | 74.4 | DeepLogo (VGG) |
| 2D Classification | FlickrLogos-32 | MAP | 73.5 | DeepLogo (AlexNet) |
| 2D Object Detection | FlickrLogos-32 | MAP | 74.4 | DeepLogo (VGG) |
| 2D Object Detection | FlickrLogos-32 | MAP | 73.5 | DeepLogo (AlexNet) |
| 16k | FlickrLogos-32 | MAP | 74.4 | DeepLogo (VGG) |
| 16k | FlickrLogos-32 | MAP | 73.5 | DeepLogo (AlexNet) |