Multimodal Large Model Joint Learning Algorithm: Reproduction Based on KubeEdge-Ianvs

examples/cloud-robotics/single_task_learning/semantic-segmentation/README.md

@@ -0,0 +1,136 @@
+
+
+---
+
+# Quick Start about Single Task Learning Semantic Segmentation
+
+Welcome to Ianvs! Ianvs is a benchmarking platform designed to evaluate the performance of distributed synergy AI solutions in accordance with recognized standards. This quick start guide will help you test your **Single Task Learning (STL)** algorithm on Ianvs. By following these streamlined steps, you can efficiently develop and benchmark your solution within minutes.
+
+### **Prerequisites**  
+Before using Ianvs, ensure that your system meets the following requirements:  
+- A single machine (a laptop or a virtual machine is sufficient; no cluster is needed)  
+- At least 2 CPUs  
+- 4GB+ of free memory (depending on the algorithm and simulation settings)  
+- 10GB+ of free disk space  
+- An internet connection for accessing GitHub, pip, etc.  
+- Python 3.6+ installed  
+
+This guide assumes you are using **Linux** with Python 3.8. If you’re on Windows, most steps will apply, but some commands and package requirements may differ.  
+
+---
+
+## Step 1. Ianvs Installation  
+
+### Clone Ianvs  
+First, set up a workspace and clone Ianvs:  
+```shell
+mkdir /ianvs
+cd /ianvs
+
+mkdir project
+cd project
+git clone https://github.com/kubeedge/ianvs.git
+```  
+
+### Install Dependencies  
+Next, install the required third-party dependencies:  
+```shell
+sudo apt-get update
+sudo apt-get install libgl1-mesa-glx -y
+python -m pip install --upgrade pip
+
+cd ianvs
+python -m pip install ./examples/resources/third_party/*
+python -m pip install -r requirements.txt
+```  
+
+### Install Ianvs  
+Finally, install Ianvs:  
+```shell
+python setup.py install
+```  
+
+---
+
+## Step 2. Dataset Preparation  
+
+### Cloud-Robotics Dataset Summary
+
+The **Cloud-Robotics Dataset** features **annotated real-world images** with **dense semantic and instance segmentation** across **30 classes** in 7 groups (e.g., vehicles, humans, nature, objects). It includes polygonal annotations, diverse daytime scenes, dynamic objects, and varying layouts. Data is provided in a JSON format, making it ideal for pixel-level semantic labeling and benchmarking vision models for robotics.
+
+Organize the dataset for STL as shown below:  
+
+```plaintext
+Dataset/
+├── 1280x760
+│   ├── gtFine
+│   │   ├── train
+│   │   ├── test
+│   │   └── val
+│   ├── rgb
+│   │   ├── train
+│   │   ├── test
+│   │   └── val
+│   └── viz
+│       ├── train
+│       ├── test
+│       └── val
+├── 2048x1024
+│   ├── gtFine
+│   │   ├── train
+│   │   ├── test
+│   │   └── val
+│   ├── rgb
+│   │   ├── train
+│   │   ├── test
+│   │   └── val
+│   └── viz
+│       ├── train
+│       ├── test
+│       └── val
+├── 640x480
+    ├── gtFine
+    │   ├── train
+    │   ├── test
+    │   └── val
+    ├── json
+    │   ├── train
+    │   ├── test
+    │   └── val
+    ├── rgb
+    │   ├── train
+    │   ├── test
+    │   └── val
+    └── viz
+        ├── train
+        ├── test
+        └── val
+```  
+
+### Dataset Preparation Command  
+```shell
+mkdir dataset
+cd dataset
+unzip dataset.zip
+```  
+
+Update the dataset's **URL address** in the `testenv.yaml` configuration file. More details can be found in the [testenv.yaml guide](https://ianvs.readthedocs.io/en/latest/guides/how-to-test-algorithms.html#step-1-test-environment-preparation).  
+
+---
+
+## Step 3. Configure Algorithm  
+
+
+Update the algorithm's **URL address** in the `algorithm.yaml` file. Refer to the [algorithm.yaml guide](https://ianvs.readthedocs.io/en/latest/guides/how-to-test-algorithms.html#step-1-test-environment-preparation) for detailed instructions.  
+
+---
+
+## Step 4. Ianvs Execution and Results  
+
+Run Ianvs for benchmarking:  
+```shell
+cd /ianvs/project
+ianvs -f examples/cloud-robotics/single_task_learning/semantic-segmentation/benchmarkingjob.yaml
+```   
+
+---

examples/cloud-robotics/single_task_learning/semantic-segmentation/benchmarkingjob.yaml

@@ -0,0 +1,70 @@
+benchmarkingjob:
+  # job name of bechmarking; string type;
+  name: "benchmarkingjob"
+  # the url address of job workspace that will reserve the output of tests; string type;
+  workspace: "/ianvs/single_task_learning/cloud"
+
+  # the url address of test environment configuration file; string type;
+  # the file format supports yaml/yml;
+  testenv: "./examples/cloud_robot/single_task_learning/testenv/testenv.yaml"
+
+  # the configuration of test object
+  test_object:
+    # test type; string type;
+    # currently the option of value is "algorithms",the others will be added in succession.
+    type: "algorithms"
+    # test algorithm configuration files; list types;
+    algorithms:
+      # algorithm name; string type;
+      - name: "rfnet_singletask_learning"
+        # the url address of test algorithm configuration file; string type;
+        # the file format supports yaml/yml
+        url: "./examples/cloud-robotics/single_task_learning/testalgorithm/rfnet/algorithm.yaml"
+
+  # the configuration of ranking leaderboard
+  rank:
+    # rank leaderboard with metric of test case's evaluation and order ; list type;
+    # the sorting priority is based on the sequence of metrics in the list from front to back;
+    sort_by: [ { "accuracy": "descend" }, { "BWT": "descend" } ]
+
+    # visualization configuration
+    visualization:
+      # mode of visualization in the leaderboard; string type;
+      # There are quite a few possible dataitems in the leaderboard. Not all of them can be shown simultaneously on the screen.
+      # In the leaderboard, we provide the "selected_only" mode for the user to configure what is shown or is not shown.
+      mode: "selected_only"
+      # method of visualization for selected dataitems; string type;
+      # currently the options of value are as follows:
+      #  1> "print_table": print selected dataitems;
+      method: "print_table"
+
+    # selected dataitem configuration
+    # The user can add his/her interested dataitems in terms of "paradigms", "modules", "hyperparameters" and "metrics",
+    # so that the selected columns will be shown.
+    selected_dataitem:
+      # currently the options of value are as follows:
+      #   1> "all": select all paradigms in the leaderboard;
+      #   2> paradigms in the leaderboard, e.g., "singletasklearning"
+      paradigms: [ "all" ]
+      # currently the options of value are as follows:
+      #   1> "all": select all modules in the leaderboard;
+      #   2> modules in the leaderboard, e.g., "basemodel"
+      modules: [ "all" ]
+      # currently the options of value are as follows:
+      #   1> "all": select all hyperparameters in the leaderboard;
+      #   2> hyperparameters in the leaderboard, e.g., "momentum"
+      hyperparameters: [ "all" ]
+      # currently the options of value are as follows:
+      #   1> "all": select all metrics in the leaderboard;
+      #   2> metrics in the leaderboard, e.g., "F1_SCORE"
+      metrics: [ "accuracy", "task_avg_acc", "BWT", "FWT"]
+
+    # model of save selected and all dataitems in workspace `./rank` ; string type;
+    # currently the options of value are as follows:
+    #  1> "selected_and_all": save selected and all dataitems;
+    #  2> "selected_only": save selected dataitems;
+    save_mode: "selected_and_all"
+
+
+
+

examples/cloud-robotics/single_task_learning/semantic-segmentation/testalgorithm/algorithm.yaml

@@ -0,0 +1,22 @@
+algorithm:
+  paradigm_type: "singletasklearning"
+  initial_model_url: "./models/model.pth"
+
+  modules:
+    - type: "basemodel"
+      name: "RFNet"
+      url: "./examples/cloudrobotics/singletask_learning_bench/testalgorithms/rfnet/basemodel.py"
+
+      hyperparameters:
+        - momentum:
+            values:
+              - 0.95
+              - 0.5
+        - learning_rate:
+            values:
+              - 0.1
+        - epochs:
+            values:
+              - 2
+
+