datamllab / autovideo Goto Github PK
View Code? Open in Web Editor NEWAutoVideo: An Automated Video Action Recognition System
Home Page: https://autoedge.ai/
License: MIT License
AutoVideo: An Automated Video Action Recognition System
Home Page: https://autoedge.ai/
License: MIT License
Hi, in your demo video and your paper you mentioned this nice and fancy ui based on Orange. But I cannot seem to find it in the code. Can you point it to me? Also how do I bring up the ui after I install it? Thank you.
when i ran
"from autovideo import extract_frames"
I get following error
"ImportError: cannot import name 'extract_frames' from 'autovideo' (/Volumes/Disk-Data/pose estimation/autovideo-main/autovideo/init.py)"
I had a problem trying to load the function "extract_frames" with from autovideo import extract_frames
In another issue I saw that the function is deprecated. My question is, is the guide available in https://towardsdatascience.com/autovideo-an-automated-video-action-recognition-system-43198beff99d no longer valid?
What is the most current example?
d3m.exceptions.StepFailedError: Step 7 for pipeline c43355b7-0e87-499f-a9f2-defc56b6713a failed
I have trained this model using fit.py on your given dataset and saved weights in the weights directory than I run produce.py these two files run smoothly.
But when I try to run recognize.py it gives me this exception.
Hi,
First of all, I'd like to congratulate about this repo, we've found this very useful. While training TSM, we've discovered that the parameter is_shift
is by default false
. Also, the import there cannot be resolved since the original make_temporal_shift
code is not integrated into this repo.
Without is_shift
enabled, does that mean that we're using a vanilla 2D Resnet50 and averaging the output of every input image in the sequence? Am I missing anything? The original contribution of TSM was this special temporal shift in the internal feature maps of any 2D CNN model.
Thanks in advance.
Hi,
I'm trying to benchmark the hmdb51 and ucf101 datasets with the pertained weights available on Google Drive. I'm unfamiliar with axolotl library and am a little confused on how to populate fitted_pipeline['runtime'] if I don't try fitting using example/fit.py. Do you have any suggestions on how to accomplish this?
Thank you,
Rohita
currently, it is working with 0.5 (Confidence Threshold) for both "normal" and "object-falling" actions. I wanted to make changes in my confidence. So how can I change the Confidence Threshold?
Minimum size of dataset is 4, I have the following hack in produce_by_path that works.
# minimum size is 4
dataset = {
'd3mIndex': [0,1,2,3],
'video': [video_name,video_name,video_name,video_name],
'label': [0,0,0,0]
}
Hi all!
I'm running into some problems with generating fitted pipelines for the different algorithms available. So I was trying to run the following command:
python3 examples/fit.py --alg tsn --pretrained --gpu 0,1 --data_dir datasets/hmdb6/ --log_path logs/tsn.txt --save_path fittted_timelines/TSN/
And I got the following output.
--> Running on the GPU
Initializing TSN with base model: resnet50.
TSN Configurations:
input_modality: RGB
num_segments: 3
new_length: 1
consensus_module: avg
dropout_ratio: 0.8Downloading: "https://download.pytorch.org/models/resnet50-0676ba61.pth" to /home/myuser/.cache/torch/hub/checkpoints/resnet50-0676ba61.pth
100%|##########| 97.8M/97.8M [00:02<00:00, 40.4MB/s]
Downloading: "https://open-mmlab.s3.ap-northeast-2.amazonaws.com/mmaction/models/kinetics400/tsn2d_kinetics400_rgb_r50_seg3_f1s1-b702e12f.pth" to /home/myuser/.cache/torch/hub/checkpoints/tsn2d_kinetics400_rgb_r50_seg3_f1s1-b702e12f.pth
Traceback (most recent call last):
File "/home/myuser/anaconda3/envs/autovideo/lib/python3.6/site-packages/d3m/runtime.py", line 1008, in _do_run_step
self._run_step(step)
File "/home/myuser/anaconda3/envs/autovideo/lib/python3.6/site-packages/d3m/runtime.py", line 998, in _run_step
self._run_primitive(step)
File "/home/myuser/anaconda3/envs/autovideo/lib/python3.6/site-packages/d3m/runtime.py", line 873, in _run_primitive
multi_call_result = self._call_primitive_method(primitive.fit_multi_produce, fit_multi_produce_arguments)
File "/home/myuser/anaconda3/envs/autovideo/lib/python3.6/site-packages/d3m/runtime.py", line 974, in _call_primitive_method
raise error
File "/home/myuser/anaconda3/envs/autovideo/lib/python3.6/site-packages/d3m/runtime.py", line 970, in _call_primitive_method
result = method(**arguments)
File "/home/myuser/anaconda3/envs/autovideo/lib/python3.6/site-packages/d3m/primitive_interfaces/base.py", line 532, in fit_multi_produce
return self._fit_multi_produce(produce_methods=produce_methods, timeout=timeout, iterations=iterations, inputs=inputs, outputs=outputs)
File "/home/myuser/anaconda3/envs/autovideo/lib/python3.6/site-packages/d3m/primitive_interfaces/base.py", line 559, in _fit_multi_produce
fit_result = self.fit(timeout=timeout, iterations=iterations)
File "/home/myuser/autovideo/autovideo/base/supervised_base.py", line 54, in fit
self._init_model(pretrained = self.hyperparams['load_pretrained'])
File "/home/myuser/autovideo/autovideo/recognition/tsn_primitive.py", line 206, in _init_model
model_data = load_state_dict_from_url(pretrained_url)
File "/home/myuser/anaconda3/envs/autovideo/lib/python3.6/site-packages/torch/hub.py", line 553, in load_state_dict_from_url
download_url_to_file(url, cached_file, hash_prefix, progress=progress)
File "/home/myuser/anaconda3/envs/autovideo/lib/python3.6/site-packages/torch/hub.py", line 419, in download_url_to_file
u = urlopen(req)
File "/home/myuser/anaconda3/envs/autovideo/lib/python3.6/urllib/request.py", line 223, in urlopen
return opener.open(url, data, timeout)
File "/home/myuser/anaconda3/envs/autovideo/lib/python3.6/urllib/request.py", line 532, in open
response = meth(req, response)
File "/home/myuser/anaconda3/envs/autovideo/lib/python3.6/urllib/request.py", line 642, in http_response
'http', request, response, code, msg, hdrs)
File "/home/myuser/anaconda3/envs/autovideo/lib/python3.6/urllib/request.py", line 570, in error
return self._call_chain(*args)
File "/home/myuser/anaconda3/envs/autovideo/lib/python3.6/urllib/request.py", line 504, in _call_chain
result = func(*args)
File "/home/myuser/anaconda3/envs/autovideo/lib/python3.6/urllib/request.py", line 650, in http_error_default
raise HTTPError(req.full_url, code, msg, hdrs, fp)
urllib.error.HTTPError: HTTP Error 403: ForbiddenThe above exception was the direct cause of the following exception:
Traceback (most recent call last):
File "examples/fit.py", line 61, in
run(args)
File "examples/fit.py", line 49, in run
pipeline=pipeline)
File "/home/myuser/autovideo/autovideo/utils/axolotl_utils.py", line 55, in fit
raise pipeline_result.error
File "/home/myuser/anaconda3/envs/autovideo/lib/python3.6/site-packages/d3m/runtime.py", line 1039, in _run
self._do_run()
File "/home/myuser/anaconda3/envs/autovideo/lib/python3.6/site-packages/d3m/runtime.py", line 1025, in _do_run
self._do_run_step(step)
File "/home/myuser/anaconda3/envs/autovideo/lib/python3.6/site-packages/d3m/runtime.py", line 1017, in _do_run_step
) from error
d3m.exceptions.StepFailedError: Step 5 for pipeline e61792eb-f54b-44ae-931c-f0f965c5e9de failed.
As you can see, I'm having problems with an Access Denied to the .pth files hosted at Amazon Cloud. Do you have any ideas on how to fix this?
works with torch==1.9.0 , torchvision==0.10.0 because torchvision has deprecated Scale in favour of Resize but d3m does not support it yet, so need to downgrade to torchvision<0.12.0 for this repo to work.
I am trying to run the given example of hmbd6 but getting error :
Traceback (most recent call last):
File "examples/fit.py", line 56, in <module>
run(args)
File "examples/fit.py", line 20, in run
from autovideo.utils import set_log_path, logger
File "/content/autovideo/autovideo/__init__.py", line 4, in <module>
from .utils import build_pipeline, fit, produce, fit_produce, produce_by_path, compute_accuracy_with_preds
File "/content/autovideo/autovideo/utils/__init__.py", line 2, in <module>
from .axolotl_utils import *
File "/content/autovideo/autovideo/utils/axolotl_utils.py", line 12, in <module>
from axolotl.backend.simple import SimpleRunner
File "/usr/local/lib/python3.7/dist-packages/axolotl/backend/simple.py", line 5, in <module>
from d3m import runtime as runtime_module
File "/usr/local/lib/python3.7/dist-packages/d3m/runtime.py", line 23, in <module>
from d3m.contrib import pipelines as contrib_pipelines
File "/usr/local/lib/python3.7/dist-packages/d3m/contrib/pipelines/__init__.py", line 13, in <module>
assert os.path.exists(NO_SPLIT_TABULAR_SPLIT_PIPELINE_PATH)
AssertionError
Running on Google colab.
Code :
!git clone https://github.com/datamllab/autovideo.git
%cd autovideo
!pip3 install -e .
!gdown --id 1nLTjp6l6UucXEy8_eOM5Zj4Q1m79OhmT
!unzip hmdb6.zip -d datasets
!python3 examples/fit.py --alg tsn --data_dir datasets/hmdb6/ --gpu "cuda"
How to resolve it?
This issue occurs with python 3.8, typing==3.7.4.3
How can we read a RTSP LINK and get prediction in this repo. i checked the code but I think RTSP link is not handling
or how can Modify the code in your action recognition repository to accept frames as input instead of a complete video file.
is this project open-source?
Or are there any documents?
A declarative, efficient, and flexible JavaScript library for building user interfaces.
๐ Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. ๐๐๐
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google โค๏ธ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.