- Update docs

README.md

@@ -38,7 +38,7 @@ pip install -r requirements.txt
 
 ### Download trained weights
 
-- Download model weights and put it in the folder `weights`. You may also need to download the weights of [DPT model]() (a rgb2depth model). The `weights` folder will look like this:
+- Download model weights and put it in the folder `weights`. You may also need to download the weights of [DPT model](https://drive.google.com/file/d/1vU4G31_T2PJv1DkA8j-MLXfMjGa7kD3L/view?usp=sharing) (a rgb2depth model). The `weights` folder will look like this:
 
 ```bash
 ├── weights
@@ -55,6 +55,9 @@ pip install -r requirements.txt
 streamlit run streamlit_apps/app.py --server.port 9113 --browser.gatherUsageStats False --server.fileWatcherType none
 ```
 
+![_](/docs/streamlit_samples/sample1_input.png)
+![_](/docs/streamlit_samples/sample1_results.png)
+
 ## Datasets
 
 ### COME15K dataset
@@ -74,6 +77,19 @@ streamlit run streamlit_apps/app.py --server.port 9113 --browser.gatherUsageStat
 }
 ```
 
+## Acknowledgements
+
+S-MultiMAE is build on top of [MultiMAE](https://github.com/EPFL-VILAB/MultiMAE). We kindly thank the authors for releasing their code.
+
+```bib
+@article{bachmann2022multimae,
+  author    = {Roman Bachmann and David Mizrahi and Andrei Atanov and Amir Zamir},
+  title     = {{MultiMAE}: Multi-modal Multi-task Masked Autoencoders},
+  booktitle = {European Conference on Computer Vision},
+  year      = {2022},
+}
+```
+
 ## References
 
 All references are cited in these files:

streamlit_apps/app_utils/image_inference.py

@@ -60,7 +60,9 @@ def image_inference(
         disabled=img_file_buffer is None,
     )
     if is_predict:
-        with st.spinner("Processing... (it takes about 1-2 minutes)"):
+        with st.spinner(
+            "Processing... (It usually takes about 30s - 1 minute per a set of salient objects)"
+        ):
             start_time = time.time()
             pred_depth, pred_sods, pred_sms = base_inference(
                 depth_model,