added multiple models and restructured repo

README.md

@@ -1,3 +1,46 @@
 ---
-license: mit
+license: GNU AGPLv3
 ---
+
+# Drawing Participation: Reblocking
+A generative AI and collaboration tool for the common good, addressing urban migration
+
+## Model Repository
+
+### About the Project
+Drawing Participation: Reblocking allows global mapping and characterization of informal settlements and provides a scalable tool to develop cadastral maps for these communities as they plan for future growth and urbanisation.
+
+It addresses the current scarcity of municipal planning data and seeks to empower literacy and collaboration in the vernacular building process. This builds a foundation for land-tenure, municipal recognition, and essential infrastructure planning, such as energy, freshwater, and sanitation.
+
+The approach automates a wide range of tasks, scaling and supporting on-site work currently manually performed on-site by surveyors and NGOs.
+
+### 
+The project was made possible by the Prototype Fund and was funded through the Federal Ministry of Education and Research from March 2024 until August 2024. Read more on the Prototype Fund's [Demo Week 15 Website](https://prototypefund.de/project/million-neighborhoods-re-blocking/) (in German).
+
+If you are interested in supporting or joining the project or simply want to find out more, please reach out via [mail(at)drawing-participation.org](mailto:mail@drawing-participation.org]). We're looking forward to hearing from you!
+
+### Broader Scope
+The software architecture for our library building-to-parcel has three main components: 
+1. We provide a Generative Ensemble Model (GEM) to produce plausible baseline parcel maps based on satellite imagery of existing building structures, 
+2. An online multi-user design interface for residents of informal neighborhoods to modify baseline parcel maps and to crowdsource parcel maps,
+3. A version control system to record and synthesize competing resident proposals to arrive at common-ground neighborhood parcel maps.
+
+### Current specific Scope
+We have so far developed the first component, the Generative Ensemble Model. 
+
+It has has five different steps:
+(1) gathering openly accessible GIS data on building footprints and parcel outlines, 
+(2) generate training data with that data on top of satellite imagery, 
+(3) train multiple image-to-image translation models (based on a pix2pix architecture), 
+(4) use a quantitative and qualitative model evaluation approach,
+and lastly,
+(5) provide a model that can synthesize the output of different models to adapt to the specific contextual characteristics of a given neighborhood
+
+### Python Libraries
+The python libraries in our [GitHub Repositories](https://github.com/scalable-design-participation-lab/drawing-participation-reblocking) allow you to perform data acquisition, model training, and model synthesis.
+In addition, here on Hugging Face you can find a set of pre-trained models, available for download and use for users with less background knowledge or computing resources.
+
+With these tools, any user can develop vector datasets based on existing building structure information.
+
+### License
+Unless otherwise described, the code in this repository and its sub-repositories is licensed under GNU AGPLv3. Please note that some modules, extensions, fonts or code herein might be licensed otherwise. If you have any questions, don't hesitate to get in touch via [email](mailto:mail@drawing-participation.org).

ensemble-model/city-classification/softmax-resnet50/softmax-ResNet50_20-ep_64-bs_10000-images_2024-09-12.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9acf47f9530a9f4d35aaa1e2c7ee6b3457d13639722d941f3396fd212697ac7f
+size 94419894

ensemble-model/image-interpolation/vq-vae/vq-vae_16-batch_80000-samples_50-epochs_2024-10-06.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:55981c5ed4d479aa5bfffa023de8c7aca3489fbd58f197100e622a1373101cd4
+size 816674

parcel-prediction/charlotte/pix2pix/us-nc-charlotte_p2x-500-net-D.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:db6a4b2af7ab17480cceae80ee835c7320d3d6da2005cc2723fd68101938a06d
+size 11089160

parcel-prediction/charlotte/pix2pix/us-nc-charlotte_p2x-500-net-G.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1050567b94adfc0aec4b1a418f2df2e26c070e9dee6b6f1578d08c046cdde733
+size 217726476

parcel-prediction/combined/pix2pix/us-pa-pittsburgh_p2x-20k-500-net-D.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2fff3a8a19e412a73b1f8d05291c3f1726dcfe7033a539541629cfc852e44af1
+size 11089160

parcel-prediction/combined/pix2pix/us-pa-pittsburgh_p2x-20k-500-net-G.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:56f097a16bcd057668fcb087df308c0be24ede2934b004ff398eb13adb81b260
+size 217726476

parcel-prediction/manhattan/pix2pix/us-ny-manhattan_p2x-500-net-D.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e5d976a0a9ec8160d54b1aad5066974a733c9401f8e83038ccbf10ca826bc5ae
+size 11089160

parcel-prediction/manhattan/pix2pix/us-ny-manhattan_p2x-500-net-G.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:750124caad6331d0eed71e0d399f4e4591aba84280724764abfa230b04d5bf88
+size 217726476

parcel-prediction/pittsburgh/pix2pix/us-pa-pittsburgh_p2x-500-net-D.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3bedf3089a341671a1384e67e793702f00b3f71b6ac76e5a88a17b29b72f84ca
+size 11089160

parcel-prediction/pittsburgh/pix2pix/us-pa-pittsburgh_p2x-500-net-G.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:feb67453f36f8858c2c4c8ee43171a7dac46fa498bb4c6a4eb691717300ceeec
+size 217726476