DeepFont: Identify Your Font from An Image

As font is one of the core design concepts, automatic font identification and similar font suggestion from an image or photo has been on the wish list of many designers. We study the Visual Font Recognition (VFR) problem, and advance the state-of-the-art remarkably by developing the DeepFont system. First of all, we build up the first available large-scale VFR dataset, named AdobeVFR, consisting of both labeled synthetic data and partially labeled real-world data. Next, to combat the domain mismatch between available training and testing data, we introduce a Convolutional Neural Network (CNN) decomposition approach, using a domain adaptation technique based on a Stacked Convolutional Auto-Encoder (SCAE) that exploits a large corpus of unlabeled real-world text images combined with synthetic data preprocessed in a specific way. Moreover, we study a novel learning-based model compression approach, in order to reduce the DeepFont model size without sacrificing its performance. The DeepFont system achieves an accuracy of higher than 80% (top-5) on our collected dataset, and also produces a good font similarity measure for font selection and suggestion. We also achieve around 6 times compression of the model without any visible loss of recognition accuracy.

PDF Abstract

Datasets


Introduced in the Paper:

AdobeVFR syn AdobeVFR real

Used in the Paper:

VFR-Wild

Results from the Paper


Task Dataset Model Metric Name Metric Value Global Rank Result Benchmark
Font Recognition AdobeVFR real DeepFont (CAE_FR) Top-1 Error Rate 28.58 # 1
Top 5 Error Rate 18.21 # 1
Top 1 Accuracy 71.42 # 1
Top 5 Accuracy 81.79 # 1
Font Recognition AdobeVFR syn DeepFont (S) Top-1 Error Rate 1.03 # 1
Top 5 Error Rate 0 # 1
Top 1 Accuracy 98.97 # 1
Top 5 Accuracy 100 # 1
Font Recognition AdobeVFR syn DeepFont (CAE_FR) Top-1 Error Rate 6.58 # 2
Top 5 Error Rate 0 # 1
Top 1 Accuracy 93.42 # 3
Top 5 Accuracy 100 # 1
Font Recognition AdobeVFR syn DeepFont (F) Top-1 Error Rate 7.4 # 3
Top 5 Error Rate 0 # 1
Top 1 Accuracy 92.6 # 4
Top 5 Accuracy 100 # 1
Font Recognition VFR-Wild DeepFont (CAE_FR) Top-1 Error Rate 38.15 # 1
Top 5 Error Rate 20.62 # 1
Top 1 Accuracy 61.85 # 1
Top 5 Accuracy 79.38 # 1

Methods


No methods listed for this paper. Add relevant methods here