-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathDockerfile
25 lines (20 loc) · 1.2 KB
/
Dockerfile
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
FROM bitxiong/tsn
# create a directory for mounting and external folder contianing the videos of interest
RUN mkdir /video_data
# copy models and executables for computing embeddings to the temporal segment networks code directory
COPY ./src/features_GPU_compute/models /app/models/
COPY ./src/features_GPU_compute/calcSig_wOF_ensemble.sh /app/scripts/
COPY ./src/features_GPU_compute/calcSig_wOF.py /app/tools/
COPY ./src/features_GPU_compute/build_wof_clips.py /app/tools/
# set environ variables
ENV PYTHONPATH=$PYTHONPATH:/app:/app/lib/caffe-action/python:/app/pyActionRecog:
ENV TSN_ROOT='/app'
ENV COMPUTE_EPS='.000003' RANDOM_SEED='73459912436'
# script to remove any spaces from names of videos - use regular expressions
# find . -type f -name "* *.xml" -exec rename "s/\s/_/g" {} \;
# # V100-1 has 6 cpu’s
#!bash scripts/calcSig_wOF_ensemble.sh tsn/Video_clips/RunningReferenceSet tsn/Video_features 6
#!bash scripts/calcSig_wOF_ensemble.sh tsn/Video_clips/RunningSearchSet tsn/Video_features 6
#!bash scripts/calcSig_wOF_ensemble.sh tsn/Video_clips/UdogYogaReferenceSet tsn/Video_features 6
#!bash scripts/calcSig_wOF_ensemble.sh tsn/Video_clips/UdogYogaSearchSet tsn/Video_features 6
CMD ["/bin/bash"]