Skip to content

Update deeplab image url and fix guides master. #1977

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
fchollet merged 4 commits into from
Oct 23, 2024
Merged

Update deeplab image url and fix guides master. #1977

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
d9897a6
add color mapping
sachinprasadhs Oct 22, 2024
4a4c3ae
model output
sachinprasadhs Oct 22, 2024
c426094
trim output
sachinprasadhs Oct 22, 2024
c63d203
remove tmp dir
sachinprasadhs Oct 22, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
Binary file added guides/img/semantic_segmentation_deeplab_v3/learning_rate_schedule.png
View file
Open in desktop
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
Binary file modified .../img/semantic_segmentation_deeplab_v3/semantic_segmentation_deeplab_v3_18_0.png
View file
Open in desktop
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
Binary file modified .../img/semantic_segmentation_deeplab_v3/semantic_segmentation_deeplab_v3_22_0.png
View file
Open in desktop
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
Binary file modified .../img/semantic_segmentation_deeplab_v3/semantic_segmentation_deeplab_v3_32_2.png
View file
Open in desktop
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
Binary file removed ...s/img/semantic_segmentation_deeplab_v3/semantic_segmentation_deeplab_v3_9_1.png
View file
Open in desktop
Binary file not shown.
Binary file added ...s/img/semantic_segmentation_deeplab_v3/semantic_segmentation_deeplab_v3_9_3.png
View file
Open in desktop
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
31 changes: 11 additions & 20 deletions guides/ipynb/keras_hub/semantic_segmentation_deeplab_v3.ipynb
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -10,7 +10,7 @@
"\n",
"**Authors:** [Sachin Prasad](https://github.com/sachinprasadhs), [Divyashree Sreepathihalli](https://github.com/divyashreepathihalli), [Ian Stenbit](https://github.com/ianstenbit)<br>\n",
"**Date created:** 2024/10/11<br>\n",
"**Last modified:** 2024/10/11<br>\n",
"**Last modified:** 2024/10/22<br>\n",
"**Description:** DeepLabV3 training and inference with KerasHub."
]
},
Expand Down Expand Up @@ -164,28 +164,25 @@
"outputs": [],
"source": [
"filepath = keras.utils.get_file(\n",
" origin=\"https://storage.googleapis.com/keras-cv/pictures/dog.jpeg\"\n",
" origin=\"https://storage.googleapis.com/keras-cv/models/paligemma/cow_beach_1.png\"\n",
")\n",
"image = keras.utils.load_img(filepath)\n",
"image = keras.utils.img_to_array(image)\n",
"image = np.array(image)\n",
"\n",
"image = preprocessor(image)\n",
"image = keras.ops.expand_dims(image, axis=0)\n",
"preds = ops.expand_dims(ops.argmax(model(image), axis=-1), axis=-1)\n",
"preds = ops.expand_dims(ops.argmax(model.predict(image), axis=-1), axis=-1)\n",
"\n",
"\n",
"def plot_segmentation(original_image, predicted_mask):\n",
" original_image = np.squeeze(original_image, axis=0)\n",
" original_image = np.clip(original_image / 255.0, 0, 1)\n",
" predicted_mask = np.squeeze(predicted_mask, axis=0)\n",
" plt.figure(figsize=(5, 5))\n",
"\n",
" plt.subplot(1, 2, 1)\n",
" plt.imshow(original_image)\n",
" plt.imshow(original_image[0] / 255)\n",
" plt.axis(\"off\")\n",
"\n",
" plt.subplot(1, 2, 2)\n",
" plt.imshow(predicted_mask, cmap=\"gray\")\n",
" plt.imshow(predicted_mask[0])\n",
" plt.axis(\"off\")\n",
"\n",
" plt.tight_layout()\n",
Expand Down Expand Up @@ -216,7 +213,7 @@
"## Download the data\n",
"\n",
"We download Pascal VOC 2012 dataset with additional annotations provided here\n",
"[Semantic contours from inverse detectors](https://www.eecs.berkeley.edu/Research/Projects/CS/vision/grouping/semantic_contours/benchmark.tgz)\n",
"[Semantic contours from inverse detectors](https://ieeexplore.ieee.org/document/6126343)\n",
"and split them into train dataset `train_ds` and `eval_ds`."
]
},
Expand Down Expand Up @@ -779,28 +776,22 @@
"source": [
"\n",
"def plot_images_masks(images, masks, pred_masks=None):\n",
" images = (images - np.min(images)) / (np.max(images) - np.min(images))\n",
" masks = (masks - np.min(masks)) / (np.max(masks) - np.min(masks))\n",
" if pred_masks is not None:\n",
" pred_masks = (pred_masks - pred_masks.min()) / (\n",
" pred_masks.max() - pred_masks.min()\n",
" )\n",
" num_images = len(images)\n",
" plt.figure(figsize=(8, 4))\n",
" rows = 3 if pred_masks is not None else 2\n",
"\n",
" for i in range(num_images):\n",
" plt.subplot(rows, num_images, i + 1)\n",
" plt.imshow(images[i])\n",
" plt.imshow(images[i] / 255)\n",
" plt.axis(\"off\")\n",
"\n",
" plt.subplot(rows, num_images, num_images + i + 1)\n",
" plt.imshow(masks[i], cmap=\"gray\")\n",
" plt.imshow(masks[i])\n",
" plt.axis(\"off\")\n",
"\n",
" if pred_masks is not None:\n",
" plt.subplot(rows, num_images, i + 1 + 2 * num_images)\n",
" plt.imshow(pred_masks[i, ..., 0], cmap=\"gray\")\n",
" plt.imshow(pred_masks[i])\n",
" plt.axis(\"off\")\n",
"\n",
" plt.show()\n",
Expand Down Expand Up @@ -1065,7 +1056,7 @@
"images, masks = next(iter(train_ds.take(1)))\n",
"images = ops.convert_to_tensor(images)\n",
"masks = ops.convert_to_tensor(masks)\n",
"preds = ops.expand_dims(ops.argmax(model(images), axis=-1), axis=-1)\n",
"preds = ops.expand_dims(ops.argmax(model.predict(images), axis=-1), axis=-1)\n",
"masks = ops.expand_dims(ops.argmax(masks, axis=-1), axis=-1)\n",
"\n",
"plot_images_masks(images, masks, preds)"
Expand Down
31 changes: 11 additions & 20 deletions guides/keras_hub/semantic_segmentation_deeplab_v3.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2,7 +2,7 @@
Title: Semantic Segmentation with KerasHub
Authors: [Sachin Prasad](https://github.com/sachinprasadhs), [Divyashree Sreepathihalli](https://github.com/divyashreepathihalli), [Ian Stenbit](https://github.com/ianstenbit)
Date created: 2024/10/11
Last modified: 2024/10/11
Last modified: 2024/10/22
Description: DeepLabV3 training and inference with KerasHub.
Accelerator: GPU
"""
Expand Down Expand Up @@ -98,28 +98,25 @@ class label such as "person", "bike", or "background" to each individual pixel
Let us visualize the results of this pretrained model
"""
filepath = keras.utils.get_file(
origin="https://storage.googleapis.com/keras-cv/pictures/dog.jpeg"
origin="https://storage.googleapis.com/keras-cv/models/paligemma/cow_beach_1.png"
)
image = keras.utils.load_img(filepath)
image = keras.utils.img_to_array(image)
image = np.array(image)

image = preprocessor(image)
image = keras.ops.expand_dims(image, axis=0)
preds = ops.expand_dims(ops.argmax(model(image), axis=-1), axis=-1)
preds = ops.expand_dims(ops.argmax(model.predict(image), axis=-1), axis=-1)


def plot_segmentation(original_image, predicted_mask):
original_image = np.squeeze(original_image, axis=0)
original_image = np.clip(original_image / 255.0, 0, 1)
predicted_mask = np.squeeze(predicted_mask, axis=0)
plt.figure(figsize=(5, 5))

plt.subplot(1, 2, 1)
plt.imshow(original_image)
plt.imshow(original_image[0] / 255)
plt.axis("off")

plt.subplot(1, 2, 2)
plt.imshow(predicted_mask, cmap="gray")
plt.imshow(predicted_mask[0])
plt.axis("off")

plt.tight_layout()
Expand All @@ -139,7 +136,7 @@ def plot_segmentation(original_image, predicted_mask):
## Download the data

We download Pascal VOC 2012 dataset with additional annotations provided here
[Semantic contours from inverse detectors](https://www.eecs.berkeley.edu/Research/Projects/CS/vision/grouping/semantic_contours/benchmark.tgz)
[Semantic contours from inverse detectors](https://ieeexplore.ieee.org/document/6126343)
and split them into train dataset `train_ds` and `eval_ds`.
"""

Expand Down Expand Up @@ -652,28 +649,22 @@ def unpackage_inputs(inputs):


def plot_images_masks(images, masks, pred_masks=None):
images = (images - np.min(images)) / (np.max(images) - np.min(images))
masks = (masks - np.min(masks)) / (np.max(masks) - np.min(masks))
if pred_masks is not None:
pred_masks = (pred_masks - pred_masks.min()) / (
pred_masks.max() - pred_masks.min()
)
num_images = len(images)
plt.figure(figsize=(8, 4))
rows = 3 if pred_masks is not None else 2

for i in range(num_images):
plt.subplot(rows, num_images, i + 1)
plt.imshow(images[i])
plt.imshow(images[i] / 255)
plt.axis("off")

plt.subplot(rows, num_images, num_images + i + 1)
plt.imshow(masks[i], cmap="gray")
plt.imshow(masks[i])
plt.axis("off")

if pred_masks is not None:
plt.subplot(rows, num_images, i + 1 + 2 * num_images)
plt.imshow(pred_masks[i, ..., 0], cmap="gray")
plt.imshow(pred_masks[i])
plt.axis("off")

plt.show()
Expand Down Expand Up @@ -839,7 +830,7 @@ def dict_to_tuple(x):
images, masks = next(iter(train_ds.take(1)))
images = ops.convert_to_tensor(images)
masks = ops.convert_to_tensor(masks)
preds = ops.expand_dims(ops.argmax(model(images), axis=-1), axis=-1)
preds = ops.expand_dims(ops.argmax(model.predict(images), axis=-1), axis=-1)
masks = ops.expand_dims(ops.argmax(masks, axis=-1), axis=-1)

plot_images_masks(images, masks, preds)
Expand Down
75 changes: 47 additions & 28 deletions guides/md/keras_hub/semantic_segmentation_deeplab_v3.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2,7 +2,7 @@

**Authors:** [Sachin Prasad](https://github.com/sachinprasadhs), [Divyashree Sreepathihalli](https://github.com/divyashreepathihalli), [Ian Stenbit](https://github.com/ianstenbit)<br>
**Date created:** 2024/10/11<br>
**Last modified:** 2024/10/11<br>
**Last modified:** 2024/10/22<br>
**Description:** DeepLabV3 training and inference with KerasHub.


Expand Down Expand Up @@ -99,33 +99,31 @@ image_converter = keras_hub.layers.DeepLabV3ImageConverter(
preprocessor = keras_hub.models.DeepLabV3ImageSegmenterPreprocessor(image_converter)
```


Let us visualize the results of this pretrained model


```python
filepath = keras.utils.get_file(
origin="https://storage.googleapis.com/keras-cv/pictures/dog.jpeg"
origin="https://storage.googleapis.com/keras-cv/models/paligemma/cow_beach_1.png"
)
image = keras.utils.load_img(filepath)
image = keras.utils.img_to_array(image)
image = np.array(image)

image = preprocessor(image)
image = keras.ops.expand_dims(image, axis=0)
preds = ops.expand_dims(ops.argmax(model(image), axis=-1), axis=-1)
preds = ops.expand_dims(ops.argmax(model.predict(image), axis=-1), axis=-1)


def plot_segmentation(original_image, predicted_mask):
original_image = np.squeeze(original_image, axis=0)
original_image = np.clip(original_image / 255.0, 0, 1)
predicted_mask = np.squeeze(predicted_mask, axis=0)
plt.figure(figsize=(5, 5))

plt.subplot(1, 2, 1)
plt.imshow(original_image)
plt.imshow(original_image[0] / 255)
plt.axis("off")

plt.subplot(1, 2, 2)
plt.imshow(predicted_mask, cmap="gray")
plt.imshow(predicted_mask[0])
plt.axis("off")

plt.tight_layout()
Expand All @@ -136,7 +134,19 @@ plot_segmentation(image, preds)
```


![png](/guides/img/semantic_segmentation_deeplab_v3/semantic_segmentation_deeplab_v3_9_1.png)
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 5s/step

<div class="k-default-codeblock">
```

```
</div>
1/1 ━━━━━━━━━━━━━━━━━━━━ 5s 5s/step




![png](/img/guides/semantic_segmentation_deeplab_v3/semantic_segmentation_deeplab_v3_9_3.png)



Expand All @@ -150,7 +160,7 @@ metric evaluation, and inference!
## Download the data

We download Pascal VOC 2012 dataset with additional annotations provided here
[Semantic contours from inverse detectors](https://www.eecs.berkeley.edu/Research/Projects/CS/vision/grouping/semantic_contours/benchmark.tgz)
[Semantic contours from inverse detectors](https://ieeexplore.ieee.org/document/6126343)
and split them into train dataset `train_ds` and `eval_ds`.


Expand Down Expand Up @@ -670,28 +680,22 @@ segmentation masks and prediction masks as input and displays them in a grid.
```python

def plot_images_masks(images, masks, pred_masks=None):
images = (images - np.min(images)) / (np.max(images) - np.min(images))
masks = (masks - np.min(masks)) / (np.max(masks) - np.min(masks))
if pred_masks is not None:
pred_masks = (pred_masks - pred_masks.min()) / (
pred_masks.max() - pred_masks.min()
)
num_images = len(images)
plt.figure(figsize=(8, 4))
rows = 3 if pred_masks is not None else 2

for i in range(num_images):
plt.subplot(rows, num_images, i + 1)
plt.imshow(images[i])
plt.imshow(images[i] / 255)
plt.axis("off")

plt.subplot(rows, num_images, num_images + i + 1)
plt.imshow(masks[i], cmap="gray")
plt.imshow(masks[i])
plt.axis("off")

if pred_masks is not None:
plt.subplot(rows, num_images, i + 1 + 2 * num_images)
plt.imshow(pred_masks[i, ..., 0], cmap="gray")
plt.imshow(pred_masks[i])
plt.axis("off")

plt.show()
Expand All @@ -702,7 +706,7 @@ plot_images_masks(batch["images"], batch["segmentation_masks"])



![png](/guides/img/semantic_segmentation_deeplab_v3/semantic_segmentation_deeplab_v3_18_0.png)
![png](/img/guides/semantic_segmentation_deeplab_v3/semantic_segmentation_deeplab_v3_18_0.png)



Expand Down Expand Up @@ -732,7 +736,7 @@ plot_images_masks(batch["images"], batch["segmentation_masks"])



![png](/guides/img/semantic_segmentation_deeplab_v3/semantic_segmentation_deeplab_v3_22_0.png)
![png](/img/guides/semantic_segmentation_deeplab_v3/semantic_segmentation_deeplab_v3_22_0.png)



Expand All @@ -753,7 +757,7 @@ dataset cardinality is important for learning rate decay because it determines
how many steps the model will train for. The initial learning rate is
proportional to 0.007 and the decay steps are 2124. This means that the learning
rate will start at `INITIAL_LR` and then decrease to zero over 2124 steps.
![png](/guides/img/semantic_segmentation_deeplab_v3_plus/learning_rate_schedule.png)
![png](/img/guides/semantic_segmentation_deeplab_v3/learning_rate_schedule.png)


```python
Expand Down Expand Up @@ -922,18 +926,20 @@ model.fit(train_ds, validation_data=eval_ds, epochs=EPOCHS)

<div class="k-default-codeblock">
```
1/Unknown 40s 40s/step - categorical_accuracy: 0.0494 - loss: 3.4081 - mean_io_u: 0.0112
1/Unknown 40s 40s/step - categorical_accuracy: 0.1191 - loss: 3.0568 - mean_io_u: 0.0118


```
</div>
2124/2124 ━━━━━━━━━━━━━━━━━━━━ 279s 113ms/step - categorical_accuracy: 0.7188 - loss: 1.1003 - mean_io_u: 0.0934 - val_categorical_accuracy: 0.8222 - val_loss: 0.5761 - val_mean_io_u: 0.3481
2124/2124 ━━━━━━━━━━━━━━━━━━━━ 281s 114ms/step - categorical_accuracy: 0.7286 - loss: 1.0707 - mean_io_u: 0.0926 - val_categorical_accuracy: 0.8199 - val_loss: 0.5900 - val_mean_io_u: 0.3265





<div class="k-default-codeblock">
```
<keras.src.callbacks.history.History at 0x7f6868c16490>
<keras.src.callbacks.history.History at 0x7fd7a897f8d0>

```
</div>
Expand All @@ -953,13 +959,26 @@ test_ds = preprocess_inputs(test_ds)
images, masks = next(iter(train_ds.take(1)))
images = ops.convert_to_tensor(images)
masks = ops.convert_to_tensor(masks)
preds = ops.expand_dims(ops.argmax(model(images), axis=-1), axis=-1)
preds = ops.expand_dims(ops.argmax(model.predict(images), axis=-1), axis=-1)
masks = ops.expand_dims(ops.argmax(masks, axis=-1), axis=-1)

plot_images_masks(images, masks, preds)
```


1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 3s/step

<div class="k-default-codeblock">
```

```
</div>
1/1 ━━━━━━━━━━━━━━━━━━━━ 3s 3s/step




![png](/guides/img/semantic_segmentation_deeplab_v3/semantic_segmentation_deeplab_v3_32_2.png)
![png](/img/guides/semantic_segmentation_deeplab_v3/semantic_segmentation_deeplab_v3_32_2.png)



Expand Down
6 changes: 3 additions & 3 deletions scripts/guides_master.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -64,11 +64,11 @@
"title": "Getting Started with KerasHub",
},
{
"path": "Semantic Segmentation with KerasHub",
"title": "semantic_segmentation_deeplab_v3",
"path": "semantic_segmentation_deeplab_v3",
"title": "Semantic Segmentation with KerasHub",
},
{
"path": "trasnformer_pretraining",
"path": "transformer_pretraining",
"title": "Pretraining a Transformer from scratch with KerasHub",
},
{
Expand Down
Loading