-
Notifications
You must be signed in to change notification settings - Fork 1
Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
Added cleaned up experimental and data processing scripts.
- Loading branch information
Showing
47 changed files
with
51,221 additions
and
1 deletion.
There are no files selected for viewing
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,96 @@ | ||
| #!/usr/bin/python | ||
| # The contents of this file are in the public domain. See LICENSE_FOR_EXAMPLE_PROGRAMS.txt | ||
| # | ||
| # This example shows how to run a CNN based face detector using dlib. The | ||
| # example loads a pretrained model and uses it to find faces in images. The | ||
| # CNN model is much more accurate than the HOG based model shown in the | ||
| # face_detector.py example, but takes much more computational power to | ||
| # run, and is meant to be executed on a GPU to attain reasonable speed. | ||
| # | ||
| # You can download the pre-trained model from: | ||
| # http://dlib.net/files/mmod_human_face_detector.dat.bz2 | ||
| # | ||
| # The examples/faces folder contains some jpg images of people. You can run | ||
| # this program on them and see the detections by executing the | ||
| # following command: | ||
| # ./cnn_face_detector.py mmod_human_face_detector.dat ../examples/faces/*.jpg | ||
| # | ||
| # | ||
| # COMPILING/INSTALLING THE DLIB PYTHON INTERFACE | ||
| # You can install dlib using the command: | ||
| # pip install dlib | ||
| # | ||
| # Alternatively, if you want to compile dlib yourself then go into the dlib | ||
| # root folder and run: | ||
| # python setup.py install | ||
| # | ||
| # Compiling dlib should work on any operating system so long as you have | ||
| # CMake installed. On Ubuntu, this can be done easily by running the | ||
| # command: | ||
| # sudo apt-get install cmake | ||
| # | ||
| # Also note that this example requires Numpy which can be installed | ||
| # via the command: | ||
| # pip install numpy | ||
|
|
||
| ''' | ||
| This detector returns a mmod_rectangles object. This object contains a list of mmod_rectangle objects. | ||
| These objects can be accessed by simply iterating over the mmod_rectangles object | ||
| The mmod_rectangle object has two member variables, a dlib.rectangle object, and a confidence score. | ||
| It is also possible to pass a list of images to the detector. | ||
| - like this: dets = cnn_face_detector([image list], upsample_num, batch_size = 128) | ||
| In this case it will return a mmod_rectangless object. | ||
| This object behaves just like a list of lists and can be iterated over. | ||
| ''' | ||
|
|
||
| import sys | ||
| import dlib | ||
| import os | ||
| import cv2 | ||
|
|
||
| print("Dlib using cuda?") | ||
| print(dlib.DLIB_USE_CUDA) | ||
|
|
||
| cnn_face_detector = dlib.cnn_face_detection_model_v1(sys.argv[1]) | ||
| outDir = sys.argv[3] | ||
| numImg = len([name for name in os.listdir(sys.argv[2]) if os.path.isfile(os.path.join(sys.argv[2], name))]) | ||
| padding = 100 | ||
|
|
||
| boundingFile = open(sys.argv[4] + "bounding-boxes.txt","w+") | ||
| saveCrop = False | ||
|
|
||
|
|
||
| for f in range(5000,numImg + 1): | ||
| number = '{0:04d}'.format(f) | ||
| filename = sys.argv[2] + "frames" + number + ".jpg" | ||
|
|
||
| print("Processing file: {}".format(f)) | ||
| img = dlib.load_rgb_image(filename) | ||
| dets = cnn_face_detector(img, 1) | ||
| h, w = img.shape[:2] | ||
|
|
||
| #print("Number of faces detected: {}".format(len(dets))) | ||
| sortedDets = sorted(dets, key=lambda a: a.confidence, reverse=True) | ||
|
|
||
| # Only keep most confident face-- we only expect one face per frame | ||
| if(len(dets) == 0): | ||
| print('No faces detected. Using last detection result.') | ||
| else: | ||
| d = sortedDets[0] | ||
|
|
||
| if (saveCrop): | ||
| y1 = max(d.rect.top() - padding, 0) | ||
| y2 = min(d.rect.bottom() + padding, h) | ||
| x1 = max(d.rect.left() - padding, 0) | ||
| x2 = min(d.rect.right() + padding, w) | ||
| cropImg = img[y1:y2, x1:x2] | ||
|
|
||
| cropImg = cv2.cvtColor(cropImg, cv2.COLOR_BGR2RGB) | ||
| cv2.imwrite(outDir + "%04d.jpg" % f,cropImg) | ||
|
|
||
| # Save detection box coordinates | ||
| boundingFile.write('%d, %d, %d, %d\n' % (d.rect.left(), d.rect.top(), d.rect.right(), d.rect.bottom())) | ||
|
|
||
| boundingFile.close() |
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,231 @@ | ||
| #!/usr/bin/env bash | ||
| # Run format: bash -i data-pipeline <Video folder> <Script folder> <Audio filename for lipgan> | ||
| # Script will: | ||
| # 1. Split video into frames | ||
| # 2. Crop around the faces in these frames, and save the bounding box coordinates to a txt file | ||
| # 3. Run 2D landmark detection on these cropped images | ||
| # 4. Save a visual of the results | ||
| # 5. Deepfake creation, with steps 1-3 repeated | ||
| # | ||
| # Assumes 6 cameras with 29.9fps mp4 video, named camera1.MP4, ..., camera6.MP4 | ||
| # | ||
| # Written by Eleanor Tursman | ||
| # Last updated 7/2020 | ||
|
|
||
| VIDEO_LOCATION=${1:-""} | ||
| SCRIPT_LOCATION=${2:-""} | ||
| AUDIO_FILENAME=${3:-""} | ||
|
|
||
| # Exit the moment a command fails | ||
| set -euo pipefail | ||
|
|
||
| cd "${VIDEO_LOCATION}" | ||
|
|
||
| ################## Convert video to frames ################## | ||
| echo "Converting video to frames..." | ||
|
|
||
| mkdir -p "${VIDEO_LOCATION}cam1-frames" | ||
| mkdir -p "${VIDEO_LOCATION}cam2-frames" | ||
| mkdir -p "${VIDEO_LOCATION}cam3-frames" | ||
| mkdir -p "${VIDEO_LOCATION}cam4-frames" | ||
| mkdir -p "${VIDEO_LOCATION}cam5-frames" | ||
| mkdir -p "${VIDEO_LOCATION}cam6-frames" | ||
|
|
||
| ffmpeg -i "camera1.MP4" -loglevel quiet -q:v 2 "${VIDEO_LOCATION}cam1-frames/frames%04d.jpg" | ||
| echo "Video 1 done." | ||
| ffmpeg -i "camera2.MP4" -loglevel quiet -q:v 2 "${VIDEO_LOCATION}cam2-frames/frames%04d.jpg" | ||
| echo "Video 2 done." | ||
| ffmpeg -i "camera3.MP4" -loglevel quiet -q:v 2 "${VIDEO_LOCATION}cam3-frames/frames%04d.jpg" | ||
| echo "Video 3 done." | ||
| ffmpeg -i "camera4.MP4" -loglevel quiet -q:v 2 "${VIDEO_LOCATION}cam4-frames/frames%04d.jpg" | ||
| echo "Video 4 done." | ||
| ffmpeg -i "camera5.MP4" -loglevel quiet -q:v 2 "${VIDEO_LOCATION}cam5-frames/frames%04d.jpg" | ||
| echo "Video 5 done." | ||
| ffmpeg -i "camera6.MP4" -loglevel quiet -q:v 2 "${VIDEO_LOCATION}cam6-frames/frames%04d.jpg" | ||
| echo "Video 6 done." | ||
|
|
||
| ################## Get face bounding boxes ################## | ||
| echo "Cropping faces..." | ||
|
|
||
| mkdir -p "${VIDEO_LOCATION}cam1-cropped" | ||
| mkdir -p "${VIDEO_LOCATION}cam2-cropped" | ||
| mkdir -p "${VIDEO_LOCATION}cam3-cropped" | ||
| mkdir -p "${VIDEO_LOCATION}cam4-cropped" | ||
| mkdir -p "${VIDEO_LOCATION}cam5-cropped" | ||
| mkdir -p "${VIDEO_LOCATION}cam6-cropped" | ||
| mkdir -p "${VIDEO_LOCATION}/bounding-boxes" | ||
|
|
||
| python "${SCRIPT_LOCATION}cnn_face_detector.py" "${SCRIPT_LOCATION}mmod_human_face_detector.dat" "${VIDEO_LOCATION}cam1-frames/" "${VIDEO_LOCATION}cam1-cropped/" "${VIDEO_LOCATION}bounding-boxes/cam1-" | ||
| echo "Video 1 done." | ||
| python "${SCRIPT_LOCATION}cnn_face_detector.py" "${SCRIPT_LOCATION}mmod_human_face_detector.dat" "${VIDEO_LOCATION}cam2-frames/" "${VIDEO_LOCATION}cam2-cropped/" "${VIDEO_LOCATION}bounding-boxes/cam2-" | ||
| echo "Video 2 done." | ||
| python "${SCRIPT_LOCATION}cnn_face_detector.py" "${SCRIPT_LOCATION}mmod_human_face_detector.dat" "${VIDEO_LOCATION}cam3-frames/" "${VIDEO_LOCATION}cam3-cropped/" "${VIDEO_LOCATION}bounding-boxes/cam3-" | ||
| echo "Video 3 done." | ||
| python "${SCRIPT_LOCATION}cnn_face_detector.py" "${SCRIPT_LOCATION}mmod_human_face_detector.dat" "${VIDEO_LOCATION}cam4-frames/" "${VIDEO_LOCATION}cam4-cropped/" "${VIDEO_LOCATION}bounding-boxes/cam4-" | ||
| echo "Video 4 done." | ||
| python "${SCRIPT_LOCATION}cnn_face_detector.py" "${SCRIPT_LOCATION}mmod_human_face_detector.dat" "${VIDEO_LOCATION}cam5-frames/" "${VIDEO_LOCATION}cam5-cropped/" "${VIDEO_LOCATION}bounding-boxes/cam5-" | ||
| echo "Video 5 done." | ||
| python "${SCRIPT_LOCATION}cnn_face_detector.py" "${SCRIPT_LOCATION}mmod_human_face_detector.dat" "${VIDEO_LOCATION}cam6-frames/" "${VIDEO_LOCATION}cam6-cropped/" "${VIDEO_LOCATION}bounding-boxes/cam6-" | ||
| echo "Video 6 done." | ||
| # | ||
| # ################## Get 2D landmarks ################## | ||
| echo "Running 2D landmark detection..." | ||
|
|
||
| mkdir -p "${VIDEO_LOCATION}cam1-landmarks" | ||
| mkdir -p "${VIDEO_LOCATION}cam2-landmarks" | ||
| mkdir -p "${VIDEO_LOCATION}cam3-landmarks" | ||
| mkdir -p "${VIDEO_LOCATION}cam4-landmarks" | ||
| mkdir -p "${VIDEO_LOCATION}cam5-landmarks" | ||
| mkdir -p "${VIDEO_LOCATION}cam6-landmarks" | ||
|
|
||
| python3 "${SCRIPT_LOCATION}detectFeatures.py" "${VIDEO_LOCATION}cam1-frames/" "${VIDEO_LOCATION}cam1-landmarks/" "${VIDEO_LOCATION}bounding-boxes/cam1-bounding-boxes.txt" | ||
| echo "Video 1 done." | ||
| python3 "${SCRIPT_LOCATION}detectFeatures.py" "${VIDEO_LOCATION}cam2-frames/" "${VIDEO_LOCATION}cam2-landmarks/" "${VIDEO_LOCATION}bounding-boxes/cam2-bounding-boxes.txt" | ||
| echo "Video 2 done." | ||
| python3 "${SCRIPT_LOCATION}detectFeatures.py" "${VIDEO_LOCATION}cam3-frames/" "${VIDEO_LOCATION}cam3-landmarks/" "${VIDEO_LOCATION}bounding-boxes/cam3-bounding-boxes.txt" | ||
| echo "Video 3 done." | ||
| python3 "${SCRIPT_LOCATION}detectFeatures.py" "${VIDEO_LOCATION}cam4-frames/" "${VIDEO_LOCATION}cam4-landmarks/" "${VIDEO_LOCATION}bounding-boxes/cam4-bounding-boxes.txt" | ||
| echo "Video 4 done." | ||
| python3 "${SCRIPT_LOCATION}detectFeatures.py" "${VIDEO_LOCATION}cam5-frames/" "${VIDEO_LOCATION}cam5-landmarks/" "${VIDEO_LOCATION}bounding-boxes/cam5-bounding-boxes.txt" | ||
| echo "Video 5 done." | ||
| python3 "${SCRIPT_LOCATION}detectFeatures.py" "${VIDEO_LOCATION}cam6-frames/" "${VIDEO_LOCATION}cam6-landmarks/" "${VIDEO_LOCATION}bounding-boxes/cam6-bounding-boxes.txt" | ||
| echo "Video 6 done." | ||
|
|
||
| ################## Visual of 2D landmarks on original input ################## | ||
| echo "Creating visual of results..." | ||
|
|
||
| mkdir -p "${VIDEO_LOCATION}/visualization" | ||
|
|
||
| python3 "${SCRIPT_LOCATION}visualize-landmarks.py" "${VIDEO_LOCATION}" | ||
| ffmpeg -i "${VIDEO_LOCATION}visualization/%04d.jpg" -loglevel quiet -vf scale=1366:768 -q:v 2 "${VIDEO_LOCATION}landmarks-visual.mp4" | ||
|
|
||
| ################## Create deepfake ################## | ||
| echo "Creating deepfake with LipGan..." | ||
|
|
||
| mkdir -p "${VIDEO_LOCATION}cam1-lipgan" | ||
| mkdir -p "${VIDEO_LOCATION}cam2-lipgan" | ||
| mkdir -p "${VIDEO_LOCATION}cam3-lipgan" | ||
| mkdir -p "${VIDEO_LOCATION}cam4-lipgan" | ||
| mkdir -p "${VIDEO_LOCATION}cam5-lipgan" | ||
| mkdir -p "${VIDEO_LOCATION}cam6-lipgan" | ||
|
|
||
| # To run lipgan on our machine, we needed to downsample & restrict the video length to the first 4000 frames | ||
| ffmpeg -i "${VIDEO_LOCATION}camera1.MP4" -ss 0.0 -frames:v 4000 -framerate 29.97 -vf scale=1280:720 -crf 0 "${VIDEO_LOCATION}cam1-lipgan/cam1-first-4k.mp4" | ||
| echo "Video 1 done" | ||
| ffmpeg -i "${VIDEO_LOCATION}camera2.MP4" -ss 0.0 -frames:v 4000 -framerate 29.97 -vf scale=1280:720 -crf 0 "${VIDEO_LOCATION}cam2-lipgan/cam2-first-4k.mp4" | ||
| echo "Video 2 done" | ||
| ffmpeg -i "${VIDEO_LOCATION}camera3.MP4" -ss 0.0 -frames:v 4000 -framerate 29.97 -vf scale=1280:720 -crf 0 "${VIDEO_LOCATION}cam3-lipgan/cam3-first-4k.mp4" | ||
| echo "Video 3 done" | ||
| ffmpeg -i "${VIDEO_LOCATION}camera4.MP4" -ss 0.0 -frames:v 4000 -framerate 29.97 -vf scale=1280:720 -crf 0 "${VIDEO_LOCATION}cam3-lipgan/cam4-first-4k.mp4" | ||
| echo "Video 4 done" | ||
| ffmpeg -i "${VIDEO_LOCATION}camera5.MP4" -ss 0.0 -frames:v 4000 -framerate 29.97 -vf scale=1280:720 -crf 0 "${VIDEO_LOCATION}cam5-lipgan/cam5-first-4k.mp4" | ||
| echo "Video 5 done" | ||
| ffmpeg -i "${VIDEO_LOCATION}camera6.MP4" -ss 0.0 -frames:v 4000 -framerate 29.97 -vf scale=1280:720 -crf 0 "${VIDEO_LOCATION}cam6-lipgan/cam6-first-4k.mp4" | ||
| echo "Video 6 done" | ||
|
|
||
| # Create new bboxes based on landmark fit | ||
| python "${SCRIPT_LOCATION}get-new-bboxes.py" 1 "${VIDEO_LOCATION}" | ||
| python "${SCRIPT_LOCATION}get-new-bboxes.py" 2 "${VIDEO_LOCATION}" | ||
| python "${SCRIPT_LOCATION}get-new-bboxes.py" 3 "${VIDEO_LOCATION}" | ||
| python "${SCRIPT_LOCATION}get-new-bboxes.py" 4 "${VIDEO_LOCATION}" | ||
| python "${SCRIPT_LOCATION}get-new-bboxes.py" 5 "${VIDEO_LOCATION}" | ||
| python "${SCRIPT_LOCATION}get-new-bboxes.py" 6 "${VIDEO_LOCATION}" | ||
|
|
||
| source ~/anaconda3/etc/profile.d/conda.sh | ||
| conda activate lipgan | ||
|
|
||
| cd "${SCRIPT_LOCATION}../LipGAN/" | ||
|
|
||
| python batch_inference.py --checkpoint_path logs/lipgan_best_residual.h5 --face "${VIDEO_LOCATION}cam1-lipgan/cam1-first-4k.mp4" --fps 29.97 --audio "audio/${AUDIO_FILENAME}.wav" --mat "audio/${AUDIO_FILENAME}.mat" --results_dir results/ --bboxes "${VIDEO_LOCATION}/bounding-boxes/cam1-lipgan-bounding-boxes.txt" | ||
| mv results/result_voice.mp4 "${VIDEO_LOCATION}cam1-lipgan/cam1-lipgan.mp4" | ||
|
|
||
| python batch_inference.py --checkpoint_path logs/lipgan_best_residual.h5 --face "${VIDEO_LOCATION}cam2-lipgan/cam2-first-4k.mp4" --fps 29.97 --audio "audio/${AUDIO_FILENAME}.wav" --mat "audio/${AUDIO_FILENAME}.mat" --results_dir results/ --bboxes "${VIDEO_LOCATION}/bounding-boxes/cam2-lipgan-bounding-boxes.txt" | ||
| mv results/result_voice.mp4 "${VIDEO_LOCATION}cam2-lipgan/cam2-lipgan.mp4" | ||
|
|
||
| python batch_inference.py --checkpoint_path logs/lipgan_best_residual.h5 --face "${VIDEO_LOCATION}cam3-lipgan/cam3-first-4k.mp4" --fps 29.97 --audio "audio/${AUDIO_FILENAME}.wav" --mat "audio/${AUDIO_FILENAME}.mat" --results_dir results/ --bboxes "${VIDEO_LOCATION}/bounding-boxes/cam3-lipgan-bounding-boxes.txt" | ||
| mv results/result_voice.mp4 "${VIDEO_LOCATION}cam3-lipgan/cam3-lipgan.mp4" | ||
|
|
||
| python batch_inference.py --checkpoint_path logs/lipgan_best_residual.h5 --face "${VIDEO_LOCATION}cam4-lipgan/cam4-first-4k.mp4" --fps 29.97 --audio "audio/${AUDIO_FILENAME}.wav" --mat "audio/${AUDIO_FILENAME}.mat" --results_dir results/ --bboxes "${VIDEO_LOCATION}/bounding-boxes/cam4-lipgan-bounding-boxes.txt" | ||
| mv results/result_voice.mp4 "${VIDEO_LOCATION}cam4-lipgan/cam4-lipgan.mp4" | ||
|
|
||
| python batch_inference.py --checkpoint_path logs/lipgan_best_residual.h5 --face "${VIDEO_LOCATION}cam5-lipgan/cam5-first-4k.mp4" --fps 29.97 --audio "audio/${AUDIO_FILENAME}.wav" --mat "audio/${AUDIO_FILENAME}.mat" --results_dir results/ --bboxes "${VIDEO_LOCATION}/bounding-boxes/cam5-lipgan-bounding-boxes.txt" | ||
| mv results/result_voice.mp4 "${VIDEO_LOCATION}cam5-lipgan/cam5-lipgan.mp4" | ||
|
|
||
| python batch_inference.py --checkpoint_path logs/lipgan_best_residual.h5 --face "${VIDEO_LOCATION}cam6-lipgan/cam6-first-4k.mp4" --fps 29.97 --audio "audio/${AUDIO_FILENAME}.wav" --mat "audio/${AUDIO_FILENAME}.mat" --results_dir results/ --bboxes "${VIDEO_LOCATION}/bounding-boxes/cam6-lipgan-bounding-boxes.txt" | ||
| mv results/result_voice.mp4 "${VIDEO_LOCATION}cam6-lipgan/cam6-lipgan.mp4" | ||
|
|
||
| conda deactivate | ||
|
|
||
| ################## Process deepfake ################## | ||
| echo "Converting deepfake video to frames..." | ||
|
|
||
| mkdir -p "${VIDEO_LOCATION}cam1-lipgan/frames" | ||
| mkdir -p "${VIDEO_LOCATION}cam2-lipgan/frames" | ||
| mkdir -p "${VIDEO_LOCATION}cam3-lipgan/frames" | ||
| mkdir -p "${VIDEO_LOCATION}cam4-lipgan/frames" | ||
| mkdir -p "${VIDEO_LOCATION}cam5-lipgan/frames" | ||
| mkdir -p "${VIDEO_LOCATION}cam6-lipgan/frames" | ||
|
|
||
| ffmpeg -i "${VIDEO_LOCATION}cam1-lipgan/cam1-lipgan.mp4" -loglevel quiet -q:v 2 "${VIDEO_LOCATION}cam1-lipgan/frames/frames%04d.jpg" | ||
| echo "Video 1 done." | ||
| ffmpeg -i "${VIDEO_LOCATION}cam2-lipgan/cam2-lipgan.mp4" -loglevel quiet -q:v 2 "${VIDEO_LOCATION}cam2-lipgan/frames/frames%04d.jpg" | ||
| echo "Video 2 done." | ||
| ffmpeg -i "${VIDEO_LOCATION}cam3-lipgan/cam3-lipgan.mp4" -loglevel quiet -q:v 2 "${VIDEO_LOCATION}cam3-lipgan/frames/frames%04d.jpg" | ||
| echo "Video 3 done." | ||
| ffmpeg -i "${VIDEO_LOCATION}cam4-lipgan/cam4-lipgan.mp4" -loglevel quiet -q:v 2 "${VIDEO_LOCATION}cam4-lipgan/frames/frames%04d.jpg" | ||
| echo "Video 4 done." | ||
| ffmpeg -i "${VIDEO_LOCATION}cam5-lipgan/cam5-lipgan.mp4" -loglevel quiet -q:v 2 "${VIDEO_LOCATION}cam5-lipgan/frames/frames%04d.jpg" | ||
| echo "Video 5 done." | ||
| ffmpeg -i "${VIDEO_LOCATION}cam6-lipgan/cam6-lipgan.mp4" -loglevel quiet -q:v 2 "${VIDEO_LOCATION}cam6-lipgan/frames/frames%04d.jpg" | ||
| echo "Video 6 done." | ||
|
|
||
| echo "Cropping faces..." | ||
|
|
||
| mkdir -p "${VIDEO_LOCATION}cam1-lipgan/cropped" | ||
| mkdir -p "${VIDEO_LOCATION}cam2-lipgan/cropped" | ||
| mkdir -p "${VIDEO_LOCATION}cam3-lipgan/cropped" | ||
| mkdir -p "${VIDEO_LOCATION}cam4-lipgan/cropped" | ||
| mkdir -p "${VIDEO_LOCATION}cam6-lipgan/cropped" | ||
|
|
||
| python "${SCRIPT_LOCATION}cnn_face_detector.py" "${SCRIPT_LOCATION}mmod_human_face_detector.dat" "${VIDEO_LOCATION}cam1-lipgan/frames/" "${VIDEO_LOCATION}cam1-lipgan/cropped/" "${VIDEO_LOCATION}bounding-boxes/cam1-post-lipgan-" | ||
| echo "Video 1 done." | ||
| python "${SCRIPT_LOCATION}cnn_face_detector.py" "${SCRIPT_LOCATION}mmod_human_face_detector.dat" "${VIDEO_LOCATION}cam2-lipgan/frames/" "${VIDEO_LOCATION}cam2-lipgan/cropped/" "${VIDEO_LOCATION}bounding-boxes/cam2-post-lipgan-" | ||
| echo "Video 2 done." | ||
| python "${SCRIPT_LOCATION}cnn_face_detector.py" "${SCRIPT_LOCATION}mmod_human_face_detector.dat" "${VIDEO_LOCATION}cam3-lipgan/frames/" "${VIDEO_LOCATION}cam3-lipgan/cropped/" "${VIDEO_LOCATION}bounding-boxes/cam3-post-lipgan-" | ||
| echo "Video 3 done." | ||
| python "${SCRIPT_LOCATION}cnn_face_detector.py" "${SCRIPT_LOCATION}mmod_human_face_detector.dat" "${VIDEO_LOCATION}cam4-lipgan/frames/" "${VIDEO_LOCATION}cam4-lipgan/cropped/" "${VIDEO_LOCATION}bounding-boxes/cam4-post-lipgan-" | ||
| echo "Video 4 done." | ||
| python "${SCRIPT_LOCATION}cnn_face_detector.py" "${SCRIPT_LOCATION}mmod_human_face_detector.dat" "${VIDEO_LOCATION}cam5-lipgan/frames/" "${VIDEO_LOCATION}cam5-lipgan/cropped/" "${VIDEO_LOCATION}bounding-boxes/cam5-post-lipgan-" | ||
| echo "Video 5 done." | ||
| python "${SCRIPT_LOCATION}cnn_face_detector.py" "${SCRIPT_LOCATION}mmod_human_face_detector.dat" "${VIDEO_LOCATION}cam6-lipgan/frames/" "${VIDEO_LOCATION}cam6-lipgan/cropped/" "${VIDEO_LOCATION}bounding-boxes/cam6-post-lipgan-" | ||
| echo "Video 6 done." | ||
|
|
||
| echo "Running 2D landmark detection..." | ||
|
|
||
| mkdir -p "${VIDEO_LOCATION}cam1-lipgan/landmarks" | ||
| mkdir -p "${VIDEO_LOCATION}cam2-lipgan/landmarks" | ||
| mkdir -p "${VIDEO_LOCATION}cam3-lipgan/landmarks" | ||
| mkdir -p "${VIDEO_LOCATION}cam4-lipgan/landmarks" | ||
| mkdir -p "${VIDEO_LOCATION}cam5-lipgan/landmarks" | ||
| mkdir -p "${VIDEO_LOCATION}cam6-lipgan/landmarks" | ||
|
|
||
| python3 "${SCRIPT_LOCATION}detectFeatures.py" "${VIDEO_LOCATION}cam1-lipgan/frames/" "${VIDEO_LOCATION}cam1-lipgan/landmarks/" "${VIDEO_LOCATION}bounding-boxes/cam1-post-lipgan-bounding-boxes.txt" | ||
| echo "Video 1 done." | ||
| python3 "${SCRIPT_LOCATION}detectFeatures.py" "${VIDEO_LOCATION}cam2-lipgan/frames/" "${VIDEO_LOCATION}cam2-lipgan/landmarks/" "${VIDEO_LOCATION}bounding-boxes/cam2-post-lipgan-bounding-boxes.txt" | ||
| echo "Video 2 done." | ||
| python3 "${SCRIPT_LOCATION}detectFeatures.py" "${VIDEO_LOCATION}cam3-lipgan/frames/" "${VIDEO_LOCATION}cam3-lipgan/landmarks/" "${VIDEO_LOCATION}bounding-boxes/cam3-post-lipgan-bounding-boxes.txt" | ||
| echo "Video 3 done." | ||
| python3 "${SCRIPT_LOCATION}detectFeatures.py" "${VIDEO_LOCATION}cam4-lipgan/frames/" "${VIDEO_LOCATION}cam4-lipgan/landmarks/" "${VIDEO_LOCATION}bounding-boxes/cam4-post-lipgan-bounding-boxes.txt" | ||
| echo "Video 4 done." | ||
| python3 "${SCRIPT_LOCATION}detectFeatures.py" "${VIDEO_LOCATION}cam5-lipgan/frames/" "${VIDEO_LOCATION}cam5-lipgan/landmarks/" "${VIDEO_LOCATION}bounding-boxes/cam5-post-lipgan-bounding-boxes.txt" | ||
| echo "Video 5 done." | ||
| python3 "${SCRIPT_LOCATION}detectFeatures.py" "${VIDEO_LOCATION}cam6-lipgan/frames/" "${VIDEO_LOCATION}cam6-lipgan/landmarks/" "${VIDEO_LOCATION}bounding-boxes/cam6-post-lipgan-bounding-boxes.txt" | ||
| echo "Video 6 done." | ||
|
|
||
|
|
||
|
|
||
|
|
||
|
|
||
|
|
||
|
|
||
|
|
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,38 @@ | ||
| # From: https://github.com/1adrianb/face-alignment | ||
|
|
||
| import face_alignment | ||
| from skimage import io | ||
| import numpy as np | ||
| import sys | ||
| import os | ||
|
|
||
| inDir = sys.argv[1] | ||
| outDir = sys.argv[2] | ||
| bounding = sys.argv[3] | ||
|
|
||
| # Uncropped image at a time, dlib detector | ||
| fa = face_alignment.FaceAlignment(face_alignment.LandmarksType._2D) | ||
|
|
||
| boundingFile = open(bounding, "r") | ||
|
|
||
| numImg = len([name for name in os.listdir(inDir) if os.path.isfile(os.path.join(inDir, name))]) | ||
| for f in range(1,numImg+1): | ||
| filename = inDir + "frames" + '{0:04d}'.format(f) + ".jpg" | ||
|
|
||
| box = boundingFile.readline().split(',') | ||
| boxInt = [[int(box[0]),int(box[1]), int(box[2]), int(box[3])]] | ||
|
|
||
| print("Processing file: {}".format(f)) | ||
| img = io.imread(filename) | ||
| pred = fa.get_landmarks_from_image(img,detected_faces=boxInt) | ||
| pred = np.array(pred) | ||
|
|
||
| if (pred.size == 1): | ||
| print("No face detected") | ||
| else: | ||
| curData = np.reshape(pred,(68,-1)) | ||
|
|
||
| np.save(outDir + 'landmarks2D-' + '{0:04d}'.format(f) + '.npy',curData) | ||
|
|
||
|
|
||
| boundingFile.close() |
Oops, something went wrong.