From 0b34ed9a012a9840b0142c0d3d58aca21d76ba4d Mon Sep 17 00:00:00 2001 From: Filip Radenovic Date: Tue, 9 Oct 2018 16:49:05 +0200 Subject: [PATCH] Updated README to incorporate the description of the sketch-based image retrieval and shape matching script. --- README.md | 9 +++++---- 1 file changed, 5 insertions(+), 4 deletions(-) diff --git a/README.md b/README.md index 20eb596..ed8285e 100644 --- a/README.md +++ b/README.md @@ -24,7 +24,7 @@ In order to run this toolbox you will need: 1. MatConvNet MATLAB toolbox version [1.0-beta25](http://www.vlfeat.org/matconvnet/download/matconvnet-1.0-beta25.tar.gz) 1. All the rest (data + networks) is automatically downloaded with our scripts -## Image retrieval (training and testing) +## Image retrieval This code implements: @@ -53,11 +53,11 @@ We provide the pretrained networks trained using the same parameters as in our E **Note**: Data and networks used for training and testing are automatically downloaded when using the example scripts. -## Sketch-based image retrieval and shape matching (training coming soon, testing available) +## Sketch-based image retrieval and shape matching This code implements: -1. Training (fine-tuning) CNN for sketch-based image retrieval and shape matching (coming soon) +1. Training (fine-tuning) CNN for sketch-based image retrieval and shape matching 1. Testing CNN sketch-based image retrieval on Flickr15k dataset Run the following script in MATLAB: @@ -65,9 +65,10 @@ Run the following script in MATLAB: ``` >> run [MATCONVNET_ROOT]/matlab/vl_setupnn; >> run [CNNIMAGERETRIEVAL_ROOT]/setup_cnnimageretrieval; +>> train_cnnsketch2imageretrieval; >> test_cnnsketch2imageretrieval; ``` -See ```[CNNIMAGERETRIEVAL_ROOT]/examples/test_sketch2cnnimageretrieval``` for additional details. +See ```[CNNIMAGERETRIEVAL_ROOT]/examples/train_cnnsketch2imageretrieval``` and ```[CNNIMAGERETRIEVAL_ROOT]/examples/test_cnnsketch2imageretrieval``` for additional details. We provide the pretrained networks trained using the same parameters as in our ECCV 2018 paper. The Flickr15k dataset used in the paper is slightly outdated compared to the latest one that is automatically downloaded when using this code (0.1 difference in mAP), so we report results here: