Skip to content
Branch: master
Find file History
keineahnung2345 and waleedka Fix visualize activations (#1211)
Fix visualize activations (Squashed 4 commits by @keineahnung2345)

# Get activations of a few sample layers
activations = model.run_graph([image], [
    ("input_image",        model.keras_model.get_layer("input_image").output)
])
leads to the error:
InvalidArgumentError: input_image:0 is both fed and fetched.

Revise the code according to https://stackoverflow.com/questions/39307108/placeholder-20-is-both-fed-and-fetched
Latest commit 6fe7db7 Jan 21, 2019

README.md

Color Splash Example

This is an example showing the use of Mask RCNN in a real application. We train the model to detect balloons only, and then we use the generated masks to keep balloons in color while changing the rest of the image to grayscale.

This blog post describes this sample in more detail.

Balloon Color Splash

Installation

From the Releases page page:

  1. Download mask_rcnn_balloon.h5. Save it in the root directory of the repo (the mask_rcnn directory).
  2. Download balloon_dataset.zip. Expand it such that it's in the path mask_rcnn/datasets/balloon/.

Apply color splash using the provided weights

Apply splash effect on an image:

python3 balloon.py splash --weights=/path/to/mask_rcnn/mask_rcnn_balloon.h5 --image=<file name or URL>

Apply splash effect on a video. Requires OpenCV 3.2+:

python3 balloon.py splash --weights=/path/to/mask_rcnn/mask_rcnn_balloon.h5 --video=<file name or URL>

Run Jupyter notebooks

Open the inspect_balloon_data.ipynb or inspect_balloon_model.ipynb Jupter notebooks. You can use these notebooks to explore the dataset and run through the detection pipelie step by step.

Train the Balloon model

Train a new model starting from pre-trained COCO weights

python3 balloon.py train --dataset=/path/to/balloon/dataset --weights=coco

Resume training a model that you had trained earlier

python3 balloon.py train --dataset=/path/to/balloon/dataset --weights=last

Train a new model starting from ImageNet weights

python3 balloon.py train --dataset=/path/to/balloon/dataset --weights=imagenet

The code in balloon.py is set to train for 3K steps (30 epochs of 100 steps each), and using a batch size of 2. Update the schedule to fit your needs.

You can’t perform that action at this time.