diff --git a/Dockerfile b/Dockerfile index 73ab8246..eda52ed3 100644 --- a/Dockerfile +++ b/Dockerfile @@ -22,7 +22,7 @@ COPY ./midas ./midas COPY ./*.py ./ # download model weights so the docker image can be used offline -RUN cd weights && {curl -OL https://github.com/AlexeyAB/MiDaS/releases/download/midas_dpt/dpt_hybrid-midas-501f0c75.pt; cd -; } +RUN cd weights && {curl -OL https://github.com/isl-org/MiDaS/releases/download/v3/dpt_hybrid-midas-501f0c75.pt; cd -; } RUN python3 run.py --model_type dpt_hybrid; exit 0 # entrypoint (dont forget to mount input and output directories) diff --git a/README.md b/README.md index 027f2143..8f3eede7 100644 --- a/README.md +++ b/README.md @@ -14,31 +14,31 @@ and our [preprint](https://arxiv.org/abs/2103.13413): MiDaS was trained on 10 datasets (ReDWeb, DIML, Movies, MegaDepth, WSVD, TartanAir, HRWSI, ApolloScape, BlendedMVS, IRS) with multi-objective optimization. -The original model that was trained on 5 datasets (`MIX 5` in the paper) can be found [here](https://github.com/intel-isl/MiDaS/releases/tag/v2). +The original model that was trained on 5 datasets (`MIX 5` in the paper) can be found [here](https://github.com/isl-org/MiDaS/releases/tag/v2). ### Changelog * [Sep 2021] Integrated to [Huggingface Spaces](https://huggingface.co/spaces) with [Gradio](https://github.com/gradio-app/gradio). See [Gradio Web Demo](https://huggingface.co/spaces/akhaliq/DPT-Large). * [Apr 2021] Released MiDaS v3.0: - New models based on [Dense Prediction Transformers](https://arxiv.org/abs/2103.13413) are on average [21% more accurate](#Accuracy) than MiDaS v2.1 - - Additional models can be found [here](https://github.com/intel-isl/DPT) + - Additional models can be found [here](https://github.com/isl-org/DPT) * [Nov 2020] Released MiDaS v2.1: - - New model that was trained on 10 datasets and is on average about [10% more accurate](#Accuracy) than [MiDaS v2.0](https://github.com/intel-isl/MiDaS/releases/tag/v2) - - New light-weight model that achieves [real-time performance](https://github.com/intel-isl/MiDaS/tree/master/mobile) on mobile platforms. - - Sample applications for [iOS](https://github.com/intel-isl/MiDaS/tree/master/mobile/ios) and [Android](https://github.com/intel-isl/MiDaS/tree/master/mobile/android) - - [ROS package](https://github.com/intel-isl/MiDaS/tree/master/ros) for easy deployment on robots + - New model that was trained on 10 datasets and is on average about [10% more accurate](#Accuracy) than [MiDaS v2.0](https://github.com/isl-org/MiDaS/releases/tag/v2) + - New light-weight model that achieves [real-time performance](https://github.com/isl-org/MiDaS/tree/master/mobile) on mobile platforms. + - Sample applications for [iOS](https://github.com/isl-org/MiDaS/tree/master/mobile/ios) and [Android](https://github.com/isl-org/MiDaS/tree/master/mobile/android) + - [ROS package](https://github.com/isl-org/MiDaS/tree/master/ros) for easy deployment on robots * [Jul 2020] Added TensorFlow and ONNX code. Added [online demo](http://35.202.76.57/). * [Dec 2019] Released new version of MiDaS - the new model is significantly more accurate and robust -* [Jul 2019] Initial release of MiDaS ([Link](https://github.com/intel-isl/MiDaS/releases/tag/v1)) +* [Jul 2019] Initial release of MiDaS ([Link](https://github.com/isl-org/MiDaS/releases/tag/v1)) ### Setup 1) Pick one or more models and download corresponding weights to the `weights` folder: -- For highest quality: [dpt_large](https://github.com/intel-isl/DPT/releases/download/1_0/dpt_large-midas-2f21e586.pt) -- For moderately less quality, but better speed on CPU and slower GPUs: [dpt_hybrid](https://github.com/intel-isl/DPT/releases/download/1_0/dpt_hybrid-midas-501f0c75.pt) -- For real-time applications on resource-constrained devices: [midas_v21_small](https://github.com/AlexeyAB/MiDaS/releases/download/midas_dpt/midas_v21_small-70d6b9c8.pt) -- Legacy convolutional model: [midas_v21](https://github.com/AlexeyAB/MiDaS/releases/download/midas_dpt/midas_v21-f6b98070.pt) +- For highest quality: [dpt_large](https://github.com/isl-org/MiDaS/releases/download/v3/dpt_large-midas-2f21e586.pt) +- For moderately less quality, but better speed on CPU and slower GPUs: [dpt_hybrid](https://github.com/isl-org/MiDaS/releases/download/v3/dpt_hybrid-midas-501f0c75.pt) +- For real-time applications on resource-constrained devices: [midas_v21_small](https://github.com/isl-org/MiDaS/releases/download/v2_1/midas_v21_small-70d6b9c8.pt) +- Legacy convolutional model: [midas_v21](https://github.com/isl-org/MiDaS/releases/download/v2_1/midas_v21-f6b98070.pt) 2) Set up dependencies: @@ -92,18 +92,18 @@ The pretrained model is also available on [PyTorch Hub](https://pytorch.org/hub/ #### via TensorFlow or ONNX -See [README](https://github.com/intel-isl/MiDaS/tree/master/tf) in the `tf` subdirectory. +See [README](https://github.com/isl-org/MiDaS/tree/master/tf) in the `tf` subdirectory. Currently only supports MiDaS v2.1. DPT-based models to be added. #### via Mobile (iOS / Android) -See [README](https://github.com/intel-isl/MiDaS/tree/master/mobile) in the `mobile` subdirectory. +See [README](https://github.com/isl-org/MiDaS/tree/master/mobile) in the `mobile` subdirectory. #### via ROS1 (Robot Operating System) -See [README](https://github.com/intel-isl/MiDaS/tree/master/ros) in the `ros` subdirectory. +See [README](https://github.com/isl-org/MiDaS/tree/master/ros) in the `ros` subdirectory. Currently only supports MiDaS v2.1. DPT-based models to be added. @@ -119,10 +119,10 @@ Zero-shot error (the lower - the better) and speed (FPS): | MiDaS v2.1 small [URL]() | 0.1344 | **0.1344** | 0.3370 | 29.27 | **13.43** | **14.53** | 30 | | | | | | | | | | **Big models:** | | | | | | | GPU RTX 3090 | -| MiDaS v2 large [URL](https://github.com/intel-isl/MiDaS/releases/download/v2/model-f46da743.pt) | 0.1246 | 0.1290 | 0.3270 | 23.90 | 9.55 | 14.29 | 51 | -| MiDaS v2.1 large [URL](https://github.com/AlexeyAB/MiDaS/releases/download/midas_dpt/midas_v21-f6b98070.pt) | 0.1295 | 0.1155 | 0.3285 | 16.08 | 8.71 | 12.51 | 51 | -| MiDaS v3.0 DPT-Hybrid [URL](https://github.com/intel-isl/DPT/releases/download/1_0/dpt_hybrid-midas-501f0c75.pt) | 0.1106 | 0.0934 | 0.2741 | 11.56 | 8.69 | 10.89 | 46 | -| MiDaS v3.0 DPT-Large [URL](https://github.com/intel-isl/DPT/releases/download/1_0/dpt_large-midas-2f21e586.pt) | **0.1082** | **0.0888** | **0.2697** | **8.46** | **8.32** | **9.97** | 47 | +| MiDaS v2 large [URL](https://github.com/isl-org/MiDaS/releases/download/v2/model-f46da743.pt) | 0.1246 | 0.1290 | 0.3270 | 23.90 | 9.55 | 14.29 | 51 | +| MiDaS v2.1 large [URL](https://github.com/isl-org/MiDaS/releases/download/v2_1/midas_v21-f6b98070.pt) | 0.1295 | 0.1155 | 0.3285 | 16.08 | 8.71 | 12.51 | 51 | +| MiDaS v3.0 DPT-Hybrid [URL](https://github.com/isl-org/MiDaS/releases/download/v3/dpt_hybrid-midas-501f0c75.pt) | 0.1106 | 0.0934 | 0.2741 | 11.56 | 8.69 | 10.89 | 46 | +| MiDaS v3.0 DPT-Large [URL](https://github.com/isl-org/MiDaS/releases/download/v3/dpt_large-midas-2f21e586.pt) | **0.1082** | **0.0888** | **0.2697** | **8.46** | **8.32** | **9.97** | 47 | diff --git a/hubconf.py b/hubconf.py index 5be7088d..99e66189 100644 --- a/hubconf.py +++ b/hubconf.py @@ -20,7 +20,7 @@ def DPT_Large(pretrained=True, **kwargs): if pretrained: checkpoint = ( - "https://github.com/intel-isl/DPT/releases/download/1_0/dpt_large-midas-2f21e586.pt" + "https://github.com/isl-org/MiDaS/releases/download/v3/dpt_large-midas-2f21e586.pt" ) state_dict = torch.hub.load_state_dict_from_url( checkpoint, map_location=torch.device('cpu'), progress=True, check_hash=True @@ -43,7 +43,7 @@ def DPT_Hybrid(pretrained=True, **kwargs): if pretrained: checkpoint = ( - "https://github.com/intel-isl/DPT/releases/download/1_0/dpt_hybrid-midas-501f0c75.pt" + "https://github.com/isl-org/MiDaS/releases/download/v3/dpt_hybrid-midas-501f0c75.pt" ) state_dict = torch.hub.load_state_dict_from_url( checkpoint, map_location=torch.device('cpu'), progress=True, check_hash=True @@ -62,7 +62,7 @@ def MiDaS(pretrained=True, **kwargs): if pretrained: checkpoint = ( - "https://github.com/intel-isl/MiDaS/releases/download/v2_1/model-f6b98070.pt" + "https://github.com/isl-org/MiDaS/releases/download/v2_1/model-f6b98070.pt" ) state_dict = torch.hub.load_state_dict_from_url( checkpoint, map_location=torch.device('cpu'), progress=True, check_hash=True @@ -81,7 +81,7 @@ def MiDaS_small(pretrained=True, **kwargs): if pretrained: checkpoint = ( - "https://github.com/intel-isl/MiDaS/releases/download/v2_1/model-small-70d6b9c8.pt" + "https://github.com/isl-org/MiDaS/releases/download/v2_1/model-small-70d6b9c8.pt" ) state_dict = torch.hub.load_state_dict_from_url( checkpoint, map_location=torch.device('cpu'), progress=True, check_hash=True diff --git a/mobile/android/README.md b/mobile/android/README.md index aecdb6b2..faf415eb 100644 --- a/mobile/android/README.md +++ b/mobile/android/README.md @@ -18,4 +18,4 @@ To use another model, you should convert it to `model_opt.tflite` and place it t ---- -Original repository: https://github.com/intel-isl/MiDaS +Original repository: https://github.com/isl-org/MiDaS diff --git a/mobile/android/models/download.gradle b/mobile/android/models/download.gradle index 0f9da676..ce76974a 100644 --- a/mobile/android/models/download.gradle +++ b/mobile/android/models/download.gradle @@ -1,4 +1,4 @@ -def modelFloatDownloadUrl = "https://github.com/intel-isl/MiDaS/releases/download/v2_1/model_opt.tflite" +def modelFloatDownloadUrl = "https://github.com/isl-org/MiDaS/releases/download/v2_1/model_opt.tflite" def modelFloatFile = "model_opt.tflite" task downloadModelFloat(type: Download) { diff --git a/mobile/ios/README.md b/mobile/ios/README.md index f430fd03..7b8eb29f 100644 --- a/mobile/ios/README.md +++ b/mobile/ios/README.md @@ -33,7 +33,7 @@ pip install tensorflow ### Install TensorFlowLiteSwift via Cocoapods -Set required TensorFlowLiteSwift version in the file (`0.0.1-nightly` is recommended): https://github.com/AlexeyAB/midas_tf_ios/blob/main/Podfile#L9 +Set required TensorFlowLiteSwift version in the file (`0.0.1-nightly` is recommended): https://github.com/isl-org/MiDaS/blob/master/mobile/ios/Podfile#L9 Install: brew, ruby, cocoapods @@ -82,7 +82,7 @@ open(model_tflite_name, "wb").write("model.tflite") ---- -Original repository: https://github.com/intel-isl/MiDaS +Original repository: https://github.com/isl-org/MiDaS ### Examples: diff --git a/mobile/ios/RunScripts/download_models.sh b/mobile/ios/RunScripts/download_models.sh index 03a4bcd0..d737b39d 100644 --- a/mobile/ios/RunScripts/download_models.sh +++ b/mobile/ios/RunScripts/download_models.sh @@ -3,7 +3,7 @@ TFLITE_MODEL="model_opt.tflite" TFLITE_FILE="Midas/Model/${TFLITE_MODEL}" -MODEL_SRC="https://github.com/intel-isl/MiDaS/releases/download/v2/${TFLITE_MODEL}" +MODEL_SRC="https://github.com/isl-org/MiDaS/releases/download/v2/${TFLITE_MODEL}" if test -f "${TFLITE_FILE}"; then echo "INFO: TF Lite model already exists. Skip downloading and use the local model." diff --git a/ros/README.md b/ros/README.md index 7da9666c..535ba121 100644 --- a/ros/README.md +++ b/ros/README.md @@ -18,14 +18,14 @@ MiDaS is a neural network to compute depth from a single image. * install ROS Melodic for Ubuntu 17.10 / 18.04: ```bash -wget https://raw.githubusercontent.com/intel-isl/MiDaS/master/ros/additions/install_ros_melodic_ubuntu_17_18.sh +wget https://raw.githubusercontent.com/isl-org/MiDaS/master/ros/additions/install_ros_melodic_ubuntu_17_18.sh ./install_ros_melodic_ubuntu_17_18.sh ``` or Noetic for Ubuntu 20.04: ```bash -wget https://raw.githubusercontent.com/intel-isl/MiDaS/master/ros/additions/install_ros_noetic_ubuntu_20.sh +wget https://raw.githubusercontent.com/isl-org/MiDaS/master/ros/additions/install_ros_noetic_ubuntu_20.sh ./install_ros_noetic_ubuntu_20.sh ``` @@ -61,7 +61,7 @@ source ~/.bashrc cd ~/ mkdir catkin_ws cd catkin_ws -git clone https://github.com/intel-isl/MiDaS +git clone https://github.com/isl-org/MiDaS mkdir src cp -r MiDaS/ros/* src diff --git a/ros/additions/downloads.sh b/ros/additions/downloads.sh index fd4b1736..9c967d4e 100644 --- a/ros/additions/downloads.sh +++ b/ros/additions/downloads.sh @@ -1,5 +1,5 @@ mkdir ~/.ros -wget https://github.com/intel-isl/MiDaS/releases/download/v2_1/model-small-traced.pt +wget https://github.com/isl-org/MiDaS/releases/download/v2_1/model-small-traced.pt cp ./model-small-traced.pt ~/.ros/model-small-traced.pt diff --git a/ros/midas_cpp/package.xml b/ros/midas_cpp/package.xml index 1b346fc1..9cac90eb 100644 --- a/ros/midas_cpp/package.xml +++ b/ros/midas_cpp/package.xml @@ -6,7 +6,7 @@ Alexey Bochkovskiy MIT - https://github.com/AlexeyAB/midas_ros + https://github.com/isl-org/MiDaS/tree/master/ros diff --git a/tf/README.md b/tf/README.md index a613da6c..5b5fe0e6 100644 --- a/tf/README.md +++ b/tf/README.md @@ -11,8 +11,8 @@ ### Run inference on TensorFlow-model by using TensorFlow -1) Download the model weights [model-f6b98070.pb](https://github.com/intel-isl/MiDaS/releases/download/v2_1/model-f6b98070.pb) -and [model-small.pb](https://github.com/intel-isl/MiDaS/releases/download/v2_1/model-small.pb) and place the +1) Download the model weights [model-f6b98070.pb](https://github.com/isl-org/MiDaS/releases/download/v2_1/model-f6b98070.pb) +and [model-small.pb](https://github.com/isl-org/MiDaS/releases/download/v2_1/model-small.pb) and place the file in the `/tf/` folder. 2) Set up dependencies: @@ -47,8 +47,8 @@ pip install -I grpcio tensorflow==2.3.0 tensorflow-addons==0.11.2 numpy==1.18.0 ### Run inference on ONNX-model by using ONNX-Runtime -1) Download the model weights [model-f6b98070.onnx](https://github.com/intel-isl/MiDaS/releases/download/v2_1/model-f6b98070.onnx) -and [model-small.onnx](https://github.com/intel-isl/MiDaS/releases/download/v2_1/model-small.onnx) and place the +1) Download the model weights [model-f6b98070.onnx](https://github.com/isl-org/MiDaS/releases/download/v2_1/model-f6b98070.onnx) +and [model-small.onnx](https://github.com/isl-org/MiDaS/releases/download/v2_1/model-small.onnx) and place the file in the `/tf/` folder. 2) Set up dependencies: @@ -87,7 +87,7 @@ pip install onnxruntime==1.5.2 ### Make ONNX model from downloaded Pytorch model file -1) Download the model weights [model-f6b98070.pt](https://github.com/intel-isl/MiDaS/releases/download/v2_1/model-f6b98070.pt) and place the +1) Download the model weights [model-f6b98070.pt](https://github.com/isl-org/MiDaS/releases/download/v2_1/model-f6b98070.pt) and place the file in the root folder. 2) Set up dependencies: