Baseline for SNAKE Challenge on AIcrowd¶
Downloads and Installs¶
In [ ]:
!wget -q https://datasets.aicrowd.com/default/aicrowd-practice-challenges/public/snake/v0.1/train.zip
!wget -q https://datasets.aicrowd.com/default/aicrowd-practice-challenges/public/snake/v0.1/val.zip
!wget -q https://datasets.aicrowd.com/default/aicrowd-practice-challenges/public/snake/v0.1/test.zip
In [ ]:
!wget -q https://datasets.aicrowd.com/default/aicrowd-practice-challenges/public/snake/v0.1/sample_submission.csv
In [ ]:
!mkdir data
!mkdir data/test
!mkdir data/train
!mkdir data/val
In [ ]:
!unzip -q train.zip
!unzip -q test.zip
!unzip -q val.zip
In [ ]:
!mv content/data/train/* data/train
!mv content/data/val/* data/val
!mv content/data/test/* data/test
In [ ]:
!rm -rf content/
In [ ]:
!mkdir models
Imports¶
In [ ]:
import sys
import os
import gc
import warnings
import torch
import torch.nn as nn
import numpy as np
import pandas as pd
import torch.nn.functional as F
from fastai.script import *
from fastai.vision import *
from fastai.callbacks import *
from fastai.distributed import *
from fastprogress import fastprogress
from torchvision.models import *
from tqdm.notebook import tqdm
In [ ]:
np.random.seed(23)
torch.cuda.device(0)
warnings.filterwarnings("ignore")
torch.multiprocessing.freeze_support()
print("[INFO] GPU:", torch.cuda.get_device_name())
Verify Images¶
In [ ]:
def verify_folder(path):
p = Path(path)
classes=['venomous','non_venomous']
for c in classes:
print(c)
verify_images(p/c, delete=True)
In [ ]:
verify_folder("data/train")
verify_folder("data/val")
In [ ]:
# verify images for test
p = Path("data/test")
verify_images(p, delete=True)
DataBunch and Model¶
Here we use a technique called progressive resizing. At each step the model is loaded with weights trained on images of lower sizes.
In [ ]:
def get_data(size, batch_size):
"""
function that returns a DataBunch as needed for the Learner
"""
tfms = get_transforms(do_flip=True, flip_vert=False, max_rotate=10.0,
max_zoom=1.1, max_lighting=0.2, max_warp=0.2, p_affine=0.75, p_lighting=0.75)
src = (ImageList.from_folder("data/train").split_none().label_from_folder())
src.valid = (ImageList.from_folder("data/val").split_none().label_from_folder()).train
src = src.transform(tfms, size=size)
src.test = (ImageList.from_folder("data/test/").split_none()).train
data = (src.databunch(num_workers=0, bs=batch_size).normalize(imagenet_stats))
print("[INFO] Number of Classes: ", data.c)
data.num_workers = 4
return data
In [ ]:
sample_data = get_data(128, (2048//128))
sample_data.show_batch(3, 3)
As you can see, the transforms have been applied and the image is normalized as well!
We first initialize all the models.
In [ ]:
learn = create_cnn(get_data(128, (2048//128)), models.resnet18,
metrics=[accuracy, FBeta(beta=1,average='macro')])
learn.model_dir = "models/"
learn.save("resnet_128")
learn = create_cnn(get_data(224, (2048//224)), models.resnet18,
metrics=[accuracy, FBeta(beta=1,average='macro')]).load("resnet_128")
learn.model_dir = "models/"
learn.save("resnet_224")
In [ ]:
def train_model(size, iter1, iter2):
"""
function to quickly train a model for a certain number of iterations.
"""
size_match = {"224": "128"}
learn = create_cnn(get_data(size, (2048//size)), models.densenet161,
metrics=[accuracy,
FBeta(beta=1,average='macro')])
learn.model_dir = "models/"
if str(size) != str(128):
learn.load("resnet_" + str(size_match[str(size)]))
name = "resnet_" + str(size)
print("[INFO] Training for : ", name)
learn.fit_one_cycle(iter1, 1e-4, callbacks=[ShowGraph(learn),
SaveModelCallback(learn, monitor='f_beta', mode='max', name=name)])
learn.unfreeze()
learn.fit_one_cycle(iter2, 5e-5, callbacks=[ShowGraph(learn),
SaveModelCallback(learn, monitor='f_beta', mode='max', name=name)])
In [ ]:
train_model(128, 2, 1)
In [ ]:
train_model(224, 2, 1)
In [ ]:
size = 224
learn = create_cnn(get_data(size, (2048//size)), models.densenet161,
metrics=[accuracy,
FBeta(beta=1,average='macro')])
learn.model_dir = "models/"
learn.load("resnet_224")
interp = ClassificationInterpretation.from_learner(learn)
losses, idxs = interp.top_losses()
display(interp.plot_top_losses(9, figsize=(15,11)))
display(interp.plot_confusion_matrix(figsize=(12,12), dpi=100))
In [ ]:
print("[INFO] MOST CONFUSED:")
interp.most_confused(min_val=5)
Out[ ]:
In [ ]:
def make_submission(learn, name):
images = []
prediction = []
probability = []
test_path = "data/test/"
test = pd.read_csv("sample_submission.csv")
test_files = os.listdir(test_path)
test_files = [x.split(".jpg")[0] for x in test_files]
test = test.loc[test["id"].isin(test_files)]
files = test["id"]
files = [x + ".jpg" for x in files]
for i in tqdm(files):
images.append(i.split(".")[0])
img = open_image(os.path.join(test_path, i))
pred_class, pred_idx, outputs = learn.predict(img)
prediction.append(pred_class.obj)
answer = pd.DataFrame({'id': images, 'class': prediction})
display(answer.head())
answer.to_csv(name, index=False)
In [ ]:
make_submission(learn, name="submission_size224.csv")
In [ ]:
try:
from google.colab import files
files.download('submission.csv')
except ImportError as e:
print("Only for Colab")
Well Done! 👍 We are all set to make a submission and see your name on leaderborad. Lets navigate to challenge page and make one.¶
Improving Further¶
- Appropriate augmnentations
- Different models like
densenet201,resnet50 - Mixed Precision training (i.e.
to_fp16()in fastai)