first commit

hw0/main.py

@@ -0,0 +1,101 @@
+import sklearn
+import numpy as np
+import cv2
+from sklearn.decomposition import PCA
+from matplotlib import pyplot as plt
+from sklearn.neighbors import KNeighborsClassifier
+from sklearn.model_selection import cross_val_score
+# For cv2.imwrite:[0,255]; 
+# cv2.imshow expects [0,1] for floating point and [0,255] for unsigned chars
+# Usage:
+# cv2.imshow("window title", array.astype('uint8'))
+# cv2.waitKey(0)
+# cv2.destroyAllWindows()*
+
+trainLst = ["./p1_data/" + str(i) + "_" + str(j) + ".png" for i in range(1,41) for j in range(1,10)]
+testLst = ["./p1_data/" + str(i) + "_10.png" for i in range(1,41)]
+
+# training set contains first 9 images of each subject
+X_train = np.array([cv2.imread(name,0) for name in trainLst])  # (360, 2576)
+y_train = np.array([i for i in range(1,41) for _ in range(9)]) # (360,)
+
+# testing set contains the last image of each subject
+X_test = np.array([cv2.imread(name,0) for name in testLst])  # (40, 2576)
+y_test = np.array([i for i in range(1,41)])  # (40,)
+
+X_train = X_train.reshape(360,-1)
+X_test = X_test.reshape(40,-1)
+
+
+# * ==============Question 1==============
+# mean face
+mean_face =  X_train.mean(axis=0)
+cv2.imwrite("./results/mean_face.png", mean_face.reshape(56,46))
+print("mean face shape: " + str(mean_face.shape)) # (2576,)
+print("X_train shape: " + str(X_train.shape)) # (360, 2576)
+
+#pca and first four eigenfaces
+pca = PCA().fit(X_train-mean_face)
+eigenfaces = pca.components_
+
+# normalize
+a = 255/(np.max(eigenfaces, axis=1) - np.min(eigenfaces, axis=1))
+a = np.expand_dims(a, -1)
+b = -a*np.expand_dims(np.min(eigenfaces,axis=1), -1)
+eigenfaces = (eigenfaces*a + b).reshape(360, 56, 46)
+
+def show_and_save(array, title, filename):
+    plt.imshow(array, cmap='gray')
+    plt.title(title, fontsize=20)
+    plt.savefig("./results/"+ filename + ".png")
+    plt.show()
+    plt.close()
+
+show_and_save(mean_face.reshape(56,46),"mean face", "mean_face")
+for i in range(4):
+   show_and_save(eigenfaces[i],"eigenface"+ str(i+1) , "eigenface_" + str(i+1))
+
+
+# * ==============Question 2==============
+img = cv2.imread("./p1_data/8_1.png",0).reshape(1,-1)
+output = pca.transform(img - mean_face)  # (1,360)
+
+reconstructs = []
+lst = [3,50,170,240,345]
+for i in lst:
+    reconstruct_result = output[:,:i] @ pca.components_[:i] + mean_face  # matrix multiplication 
+    reconstructs.append(reconstruct_result.reshape(56,46))
+
+for i in range(5):
+    show_and_save(reconstructs[i], "n = " + str(lst[i]), "reconstruct_" + str(i))
+
+
+# * ====Question 3====
+# print(reconstructs[0].shape)  # (56,46)
+# print(img.shape)  # (1,2576)
+for i in range(len(lst)):
+    mse = np.mean((img - reconstructs[i].reshape(1,-1))**2)  # pixel-wise subtraction  
+    print("n:{:<3d}, MSE:{:<15f}".format(lst[i], mse))
+
+
+# * ====Question 4====
+kLst = [1,3,5]
+nLst = [3,50,170]
+train_pca = pca.transform(X_train - mean_face)
+#print(train_pca.shape) 
+#print(y_train.shape) # (360,)
+for k in kLst:
+    for n in nLst:
+        clf = KNeighborsClassifier(n_neighbors=k)
+        scores = cross_val_score(clf, train_pca[:,:n], y_train, cv=3, scoring="accuracy")
+        print("k={}, n={:<3d}, Acc:{:<15f}".format(k,n,scores.mean()))
+
+
+# * ====Question 5====
+k, n  = 1, 50
+test_pca = pca.transform(X_test - mean_face)
+clf = KNeighborsClassifier(n_neighbors=k)
+clf.fit(train_pca[:,:n], y_train)
+clf.score(test_pca[:,:n], y_test)
+
+print("Score:", clf.score(test_pca[:,:n], y_test))

hw1/.gitignore

@@ -0,0 +1,16 @@
+#Ignore the dataset
+#IMPORTANT: make sure that your dataset is stored in a folder with the same as the one specofied below
+hw1_data/*
+*.png
+src/log/*
+src2/log/*
+src/__pycache__/*
+src2/__pycache__/*
+*.pyc
+results/*
+INTERRUPTED.pth
+model_best_*
+src2/results/*
+src2/checkpoints/*
+src2/fcn8s-heavy-pascal.pth
+*.pth

hw1/README.md

@@ -0,0 +1,90 @@
+
+# HW1 Problem 1― Image Classification
+In HW1 problem 1, you will need to implement an image classification model and answer some questions in the report.
+
+# HW1 Problem 2― Semantic Segmentation
+In HW1 problem 2, you will need to implement two semantic segmentation models and answer some questions in the report.
+
+For more details, please click [this link](https://docs.google.com/presentation/d/1H4O5NrEK-AzS2jRggWSpzs9hHEsYlCNYQdJbHlDfH48/edit?usp=sharing) to view the slides of HW1
+
+# Usage
+To start working on this assignment, you should clone this repository into your local machine by using the following command.
+
+    git clone https://github.com/DLCV-Fall-2021/hw1-<username>.git
+Note that you should replace `<username>` with your own GitHub username.
+
+### Dataset
+In the starter code of this repository, we have provided a shell script for downloading and extracting the dataset for this assignment. For Linux users, simply use the following command.
+
+    bash ./get_dataset.sh
+The shell script will automatically download the dataset and store the data in a folder called `hw1_data`. Note that this command by default only works on Linux. If you are using other operating systems or you can not download the dataset by running the command above, you should download the dataset from [this link](https://drive.google.com/file/d/1CIvfO8rDMq5-3vmi0c6mrhDfzqrAZDem/view?usp=sharing) and unzip the compressed file manually.
+
+> ⚠️ ***IMPORTANT NOTE*** ⚠️  
+> You should keep a copy of the dataset only in your local machine. **DO NOT** upload the dataset to this remote repository. If you extract the dataset manually, be sure to put them in a folder called `hw1_data` under the root directory of your local repository so that it will be included in the default `.gitignore` file.
+
+### Evaluation
+To evaluate your semantic segmentation model, you can run the provided evaluation script provided in the starter code by using the following command.
+
+    python3 mean_iou_evaluate.py <--pred PredictionDir> <--labels GroundTruthDir>
+
+ - `<PredictionDir>` should be the directory to your predicted semantic segmentation map (e.g. `hw1_data/prediction/`)
+ - `<GroundTruthDir>` should be the directory of ground truth (e.g. `hw1_data/validation/`)
+
+Note that your predicted segmentation semantic map file should have the same filename as that of its corresponding ground truth label file (both of extension ``.png``).
+
+### Visualization
+To visualization the ground truth or predicted semantic segmentation map in an image, you can run the provided visualization script provided in the starter code by using the following command.
+
+    python3 viz_mask.py <--img_path xxxx_sat.jpg> <--seg_path xxxx_mask.png>
+
+# Submission Rules
+### Deadline
+2021/10/26 (Tue.) 11:59 PM
+
+### Late Submission Policy
+You have a three-day delay quota for the whole semester. Once you have exceeded your quota, the credit of any late submission will be deducted by 30% each day.
+
+Note that while it is possible to continue your work in this repository after the deadline, **we will by default grade your last commit before the deadline** specified above. If you wish to use your quota or submit an earlier version of your repository, please contact the TAs and let them know which commit to grade.
+
+### Academic Honesty
+-   Taking any unfair advantages over other class members (or letting anyone do so) is strictly prohibited. Violating university policy would result in an **F** grade for this course (**NOT** negotiable).    
+-   If you refer to some parts of the public code, you are required to specify the references in your report (e.g. URL to GitHub repositories).      
+-   You are encouraged to discuss homework assignments with your fellow class members, but you must complete the assignment by yourself. TAs will compare the similarity of everyone’s submission. Any form of cheating or plagiarism will not be tolerated and will also result in an **F** grade for students with such misconduct.
+
+
+### Submission Format
+Aside from your own Python scripts and model files, you should make sure that your submission includes *at least* the following files in the root directory of this repository:
+ 1.   `hw1_<StudentID>.pdf`  
+The report of your homework assignment. Refer to the "*Submission*" section in the slides for what you should include in the report. Note that you should replace `<StudentID>` with your student ID, **NOT** your GitHub username.
+ 2.   `hw1_1.sh`  
+The shell script file for running your classification model.
+ 3.   `hw1_2.sh`  
+The shell script file for running your semantic segmentation model.
+
+We will run your code in the following manner:
+
+    bash hw1_1.sh $1 $2
+where `$1` is the testing images directory (e.g. `test/images/`), and `$2` is the path of folder where you want to output your prediction file (e.g. `test/label_pred/` ). Please do not create the output prediction directory in your bash script or python codes.
+
+    bash hw1_2.sh $1 $2
+where `$1` is the testing images directory (e.g. `test/images/`), and `$2` is the output prediction directory for segmentation maps (e.g. `test/label_pred/` ). Please do not create the output prediction directory in your bash script or python codes.
+
+### Packages
+This homework should be done using python3.6. For a list of packages you are allowed to import in this assignment, please refer to the requirments.txt for more details.
+
+You can run the following command to install all the packages listed in the requirements.txt:
+
+    pip3 install -r requirements.txt
+
+Note that using packages with different versions will very likely lead to compatibility issues, so make sure that you install the correct version if one is specified above. E-mail or ask the TAs first if you want to import other packages.
+
+### Remarks
+- If your model is larger than GitHub’s maximum capacity (100MB), you can upload your model to another cloud service (e.g. Dropbox). However, your shell script files should be able to download the model automatically. For a tutorial on how to do this using Dropbox, please click [this link](https://goo.gl/XvCaLR).
+- **DO NOT** hard code any path in your file or script, and the execution time of your testing code should not exceed an allowed maximum of **10 minutes**.
+- **Please refer to HW1 slides for details about the penalty that may incur if we fail to run your code or reproduce your results.**
+
+# Q&A
+If you have any problems related to HW1, you may
+- Use TA hours
+- Contact TAs by e-mail ([[email protected]](mailto:[email protected]))
+- Post your question under hw1 FAQ section in FB group

hw1/get_dataset.sh

@@ -0,0 +1,9 @@
+# Download dataset
+wget --load-cookies /tmp/cookies.txt "https://docs.google.com/uc?export=download&confirm=$(wget --quiet --save-cookies /tmp/cookies.txt --keep-session-cookies --no-check-certificate 'https://docs.google.com/uc?export=download&id=1E_X2X-91SVsv0u_b-57wOc_CZuxBKkvJ' -O- | sed -rn 's/.*confirm=([0-9A-Za-z_]+).*/\1\n/p')&id=1E_X2X-91SVsv0u_b-57wOc_CZuxBKkvJ" -O hw1_data.zip && rm -rf /tmp/cookies.txt
+
+# Unzip the downloaded zip file
+mkdir hw1_data
+unzip ./hw1_data.zip -d hw1_data
+
+# Remove the downloaded zip file
+rm ./hw1_data.zip

hw1/hw1_1.sh

@@ -0,0 +1,3 @@
+# bash hw1_1.sh $1(testing images directory) $2path of output csv file(predicted labels)
+
+python3 ./src/main.py --mode=test --test=./src/model.pth --test_data=$1 --prediction=$2

hw1/hw1_2.sh

@@ -0,0 +1,3 @@
+# hw1_2.sh $1(testing images directory) $2(output images directory)
+wget "https://www.dropbox.com/s/ult694hblco4lra/model_unet_699.pth?dl=1" -O ./src2/model_unet_699.pth
+python3 src2/main.py --mode=test --test=src2/model_unet_699.pth --model=unet --test_data=$1 --result_dir=$2