app

app.py

@@ -69,40 +69,47 @@ def predict(img_input):
     return blend_images(img_pil, T.ToPILImage()(preds_up))
     
 
-title = 'FOUND - unsupervised object localization \n CVPR23'
+title = 'FOUND - unsupervised object localization'
 description = 'Gradio Demo for our CVPR23 paper "Unsupervised Object Localization: Observing the Background to Discover Objects"\n \
     The app is <i>running on CPUs</i>, inference times are therefore longer than those expected on GPU (80 FPS on a V100 GPU).\n \
     Please see below for more details.'
     
-article = """<h1 align="center">Unsupervised Object Localization: Observing the Background to Discover Objects </h1>
+article = """
+<h1 align="center">Unsupervised Object Localization: Observing the Background to Discover Objects</h1>
     
-    ## Highlights
-
-    - Single **conv 1 x 1** layer trained to extract information from DINO [1] features.
-    - Trained only for 2 epochs on the dataset DUTS-TR.
-    - Inference runs at 80 FPS on a V100 GPU.
-    - No post-processing applied in results here.
-
-    <i> Images provided are taken from VOC07 [2], ECSSD [3] and DUT-OMRON [4].</i>
-
-    ## Citation
-    ```
-    @inproceedings{simeoni2023found,
-    author    = {Siméoni, Oriane and Sekkat, Chloé and Puy, Gilles and Vobecky, Antonin and Zablocki, Éloi and Pérez, Patrick},
-    title     = {Unsupervised Object Localization: Observing the Background to Discover Objects},
-    booktitle = {{IEEE} Conference on Computer Vision and Pattern Recognition, {CVPR}},
-    year      = {2023},
-    }
-    ```
-
-    ### References
-
-    [1] M. Caron et al. Emerging properties in self-supervised vision transformers, ICCV 2021
-    [2] M. Everingham et al. The PASCAL Visual Object Classes Challenge 2007 (VOC2007) Results
-    [3] J. Shi et al. Hierarchical image saliency detection on extended CSSD, IEEE TPAMI 2016
-    [4] C. Yang et al. Saliency detection via graph-based manifold ranking, CVPR 2013
-    """
-examples = ["data/examples/VOC_000030.jpg"]
+## Highlights
+- Single **conv 1 x 1** layer trained to extract information from DINO [1] features.
+- **No supervision**.
+- Trained only for **2 epochs** on the dataset DUTS-TR.
+- Inference runs at **80 FPS** on a V100 GPU.
+- No post-processing applied in results here.
+
+<i> Images provided are taken from VOC07 [2], ECSSD [3] and DUT-OMRON [4].</i>
+
+## Citation
+```
+@inproceedings{simeoni2023found,
+author    = {Siméoni, Oriane and Sekkat, Chloé and Puy, Gilles and Vobecky, Antonin and Zablocki, Éloi and Pérez, Patrick},
+title     = {Unsupervised Object Localization: Observing the Background to Discover Objects},
+booktitle = {{IEEE} Conference on Computer Vision and Pattern Recognition, {CVPR}},
+year      = {2023},
+}
+```
+
+### References
+
+[1] M. Caron et al. Emerging properties in self-supervised vision transformers, ICCV 2021
+[2] M. Everingham et al. The PASCAL Visual Object Classes Challenge 2007 (VOC2007) Results
+[3] J. Shi et al. Hierarchical image saliency detection on extended CSSD, IEEE TPAMI 2016
+[4] C. Yang et al. Saliency detection via graph-based manifold ranking, CVPR 2013
+
+"""
+
+examples = ["data/examples/VOC_000030.jpg",
+            "data/examples/ECSSD_0010.png",
+            "data/examples/ECSSD_0022.png",
+            "data/examples/DUT-OMRON_im103.png",
+            ]
 
 
 iface = gr.Interface(fn=predict,