Wiki: Lesson 2


#16

Thanks! So side_on means flipping horizontally, and top_down means flipping vertically? Is that right?


(ecdrid) #17

Sounds perfect…
It can be verified on a image if you want…
Just collect the pixel values in the left and right by dividing them in between and reverse them to create a new image and then plot them…

I guess this will do what we want…


(Greg Fetisov) #18

Hi to everyone! Please, help me with some error with notebook lesson2 “Multi-label classification”.

I am using a paperspace, so run cell

# Data preparation steps if you are using Crestle:
os.makedirs('data/planet/models', exist_ok=True)
os.makedirs('/cache/planet/tmp', exist_ok=True)
!ln -s /datasets/kaggle/planet-understanding-the-amazon-from-space/train-jpg {PATH}
!ln -s /datasets/kaggle/planet-understanding-the-amazon-from-space/train_v2.csv {PATH}
!ln -s /cache/planet/tmp {PATH}

Here is error output.

Any ideas? Thank you in advance!


(Aditya Bansal) #19

@GregFet
if you’re not using crestle, you dont need to run that block…you can safely ignore it :slight_smile:


#20

Thanks for your patience, got it !


(Dave Luo) #21

@ibunny, @ecdrid

You’re correct that the transforms_side_on flips the image left and right.

However, transforms_top_down is more than just vertical flipping. It’s vertical flips + horizontal flips + every possible 90-degree rotation.

I believe the naming comes from the idea that some images you would capture from the side (like taking a photo of a cat or dog) vs some you take top-down (like satellite images, or food photos on instagram…). In the side-on case, realistic data augmentations would be flipping horizontally (except in the occasional case of the sidewise or upside-down hanging cat/dog…). In top-down imaging like with satellites, you can rotate and flip the image in every direction and it could still look like a plausible training image.

Here are some examples generated using the transform functions with cat/dog lesson1 images:

original cat image:

orig_cat2

transforms_side_on, 12 examples:

side_on_cat

transforms_top_down, 12 examples (note the mirror images + rotations):

top_down_cat

Here’s a look at transforms.py:

transforms_basic    = [RandomRotateXY(10), RandomLightingXY(0.05, 0.05)]
transforms_side_on  = transforms_basic + [RandomFlipXY()]
transforms_top_down = transforms_basic + [RandomDihedralXY()]

class RandomDihedralXY(CoordTransform):
  def set_state(self):
    self.rot_times = random.randint(0,4)
    self.do_flip = random.random()<0.5

  def do_transform(self, x):
    x = np.rot90(x, self.rot_times)
    return np.fliplr(x).copy() if self.do_flip else x

class RandomFlipXY(CoordTransform):
  def set_state(self):
    self.do_flip = random.random()<0.5

  def do_transform(self, x):
    return np.fliplr(x).copy() if self.do_flip else x

Note that with both settings, there’s a bit of slight rotation and brightness adjustments included by default as well.


Wiki: Lesson 1
(ecdrid) #22

Thanks a lot @daveluo
Can you go further as to how you got those plots(insights)?
By calling the transformation functions and then plotting?
Thanks…


#23

wow, thanks a million! That’s pretty straight forward:heart_eyes:


(Dave Luo) #24

Yup, exactly. Slightly adjusted lesson1.py code in the augmentation section.

I changed the index from [1] to [0] to show a different cat than the original notebook. For the augmented images, switched aug_tfms=transforms_side_on and aug_tfms=transforms_top_down and the range/num of rows:

def get_augs():
    data = ImageClassifierData.from_paths(PATH, bs=2, tfms=tfms, num_workers=1)
    x,_ = next(iter(data.aug_dl))
    return data.trn_ds.denorm(x)[0]

tfms = tfms_from_model(resnet34, sz, aug_tfms=transforms_side_on, max_zoom=1.0)
data = ImageClassifierData.from_paths(PATH, bs=2, tfms=tfms, num_workers=1)
x,_ = next(iter(data.aug_dl))
ims = np.stack([get_augs() for i in range(12)])
plots(ims, rows=3)

For the original photo, removed all transforms:

tfms = tfms_from_model(resnet34, sz, aug_tfms=[])
data = ImageClassifierData.from_paths(PATH, bs=2, tfms=tfms, num_workers=1)
plt.imshow(data.trn_ds.denorm(x)[0])

(ecdrid) #25

Thanks a lot…
Makes things transparent…


(Greg Fetisov) #26

@adbansal

It seems obvious, but:

  1. there is a similar cell code in lessons1 (‘run if using Crestle’), but @jeremy in video said to run it if using crestle/paperspace. So I think it is rule for all non-amazon users.
  2. without this cell I have no satellite dataset on my paperspace, so this cell also useful for me.

I can download ds from kaggle, just ask here if someone knows what to do with that error. Maybe source of this error affects on other things and I should fix it. But how


(ecdrid) #27

It seems that the problem is related with access…
Try creating a directory with sudo permissions or remove the cache dir…
And recreate one…

Warning I have not tested this


(Linda Liu) #28

I have a question about the steps we take to train model:

  1. Enable data augmentation, and precompute=True
  2. Use lr_find() to find highest learning rate where loss is still clearly improving
  3. Train last layer from precomputed activations for 1-2 epochs
  4. Train last layer with data augmentation (i.e. precompute=False) for 2-3 epochs with cycle_len=1
  5. Unfreeze all layers
  6. Set earlier layers to 3x-10x lower learning rate than next higher layer
  7. Use lr_find() again
  8. Train full network with cycle_mult=2 until over-fitting

Why do we need to train the last layer in step 3 if we have to retrain the last layer again in the next step? wouldn’t the retrain change all the activation?


(Alan O'Donnell) #29

@lindarrrliu I think your #1 (or maybe the lecture’s #1) has a typo: turning on data augmentation means precompute=False (as in step #4). I think step #1 should be disable data augmentation but enable precompute.

At step #3, we’re actually only training the last layer. The idea is that the pre-trained network feeds into our little final adapter layer; we’re adapting a pretrained network in the first place because we think the features it has already learned will be a good fit for our problem. This lets us get the final layer into more or less the right place.

Once we’ve done so, we can then gently thaw the pre-trained network (with those differential learning rates) and let it get to better place too. It’s true that at this point we’re also retraining the final layer, but I think the idea is that because everything is more or less good to go already, training won’t wreck the weights we found in steps #3-#4 (assuming we haven’t goofed and reset our learning rate too high, etc.). And for that matter, as soon as you start training some layer in the middle of the network, you need to train everything afterwards too. The middle layer will start spitting out new kinds of activations, and everyone downstream will need to learn to adapt.


(Linda Liu) #30

Thanks @cqfd! You explained well :slight_smile:
Just rewatched week 3 video and I think I understand it more intuitively now. Before step 3, the weight of our final layer is at random. Step 3 and 4 update the weight of the final layer to a state that’s better than random. Than when we unfreeze all other layers, we further tune the last layer to adapt the change in earlier layers.

I wonder whether we could get to more or less the same place if we start out with random final layer and train all layers with lots iterations.


#31

@jeremy

Question: in the lesson 1 notebook, it says “we simply keep increasing the learning rate from a very small value, until the loss starts decreasing”. Shouldn’t this be until the loss starts INCREASING? Since we are trying to minimize loss?


(corey) #32

Does anybody know what the parameter ps is in this method learn = ConvLearner.pretrained(arch,data, precompute=True, ps=0.5)

I am also just generally having a hard time with learning about the fast.ai methods when I want to do something different since it is new and seemingly not well documented yet. Is there a place to look other than pressing shift + tab to look for more detail?


(ecdrid) #33

It’s from Dropouts

Too look for more detail try ??Convlearner.pretained() in a cell…


(Ian) #34

@Dhruv, apologies, I must be blind because I still don’t see it :frowning:

What is the name of the python notebook that contains the dogbreed code?


(Reshama Shaikh) #35

@corey
I’ve been keeping a list of the fastai terms. Whenever I can’t remember what something stands for, I go to the top of my repo fastai_deeplearn_part1 and do a search.

In this case, ps means p’s (plural, I believe) to represent the probability of dropout.