Home

Introduction to Pattern Recognition
 

Exam deadline

posted 22 Jun 2017, 07:51 by Simone Grazioso

I want to remember that, as prof. Pirri said during the course, the deadline for the project is the 31st of September.
Contact me if you need to discuss about the projects.

Exercitation 7

posted 16 May 2017, 01:19 by Simone Grazioso   [ updated 16 May 2017, 05:00 ]

The pdf and the code of the Exercitation 7 is available on Teaching Material section.

Exercitation 16/05/2017

posted 15 May 2017, 01:45 by Simone Grazioso

Tomorrow exercitation will be on Feature Extraction with Python Caffe. Please bring your laptop, and download/make Caffe BVLC before the lesson. Please also download the pretrained model (you can download it from here), as you know sapienza wifi is not reliable.
The exercitation will be on writing a python script that extract features vectors from various layers of the ConvNet.

Exercitation 6

posted 12 May 2017, 07:24 by Simone Grazioso

The pdf of the Exercitation 6 is available on Teaching Material section.

Exercitation 09/05/2017

posted 9 May 2017, 01:30 by Simone Grazioso   [ updated 9 May 2017, 01:33 ]

Today's exercitation will take place in the ALCOR Lab. You will be divided in three groups, and each group will have the access to the lab for a time slot (about 30min). 
I will come in classroom to make groups and take you to the lab.

Exercitation 5

posted 4 Apr 2017, 05:46 by Simone Grazioso   [ updated 9 May 2017, 01:32 ]

The pdf of the Exercitation 5 is available on Teaching Material section.

Recommended readings:
Ren, Shaoqing, et al. "Faster r-cnn: Towards real-time object detection with region proposal networks." Advances in neural information processing systems. 2015.

Extra Material
https://github.com/mdouskos/py-faster-rcnn (fork of the original repository https://github.com/rbgirshick/py-faster-rcnn)

Exercitation 04/04/2017

posted 3 Apr 2017, 07:00 by Simone Grazioso

Tomorrow's exercitation will take place in the ALCOR Lab. Given a restricted environment, you will be divided in a few groups, and each group will have the access to the lab for a time slot (depending of the number of participants).

Anyway, I will come to your classroom to make groups and take you to the lab.

Exercitation 4

posted 29 Mar 2017, 03:08 by Simone Grazioso

The slides of the Exercitation 4 are available on Teaching Material section.

Recommended readings:
Ren, Shaoqing, et al. "Faster r-cnn: Towards real-time object detection with region proposal networks." Advances in neural information processing systems. 2015.
Deng, Jia, et al. "Imagenet: A large-scale hierarchical image database." CVPR 2016
Lin, Tsung-Yi, et al. "Microsoft coco: Common objects in context." ECCV, 2014.

Extra Material
Faster R-CNN repository: https://github.com/ShaoqingRen/faster_rcnn (matlab) https://github.com/rbgirshick/py-faster-rcnn (python)
http://www.mscoco.org
http://www.image-net.org

Exercitation 3

posted 21 Mar 2017, 08:30 by Simone Grazioso   [ updated 21 Mar 2017, 08:32 ]

The slides of the Exercitation 3 are available on Teaching Material section.

Recommended readings:
Cao, Zhe, et al. "Realtime Multi-Person 2D Pose Estimation using Part Affinity Fields." arXiv preprint arXiv:1611.08050 (2016).
Girshick, Ross, et al. "Rich feature hierarchies for accurate object detection and semantic segmentation." CVPR 2014.
Girshick, Ross. "Fast r-cnn." ICCV 2015
Ren, Shaoqing, et al. "Faster r-cnn: Towards real-time object detection with region proposal networks." Advances in neural information processing systems. 2015.

Extra Material
Caffe DeepNet Framework
Faster R-CNN repository: https://github.com/ShaoqingRen/faster_rcnn (matlab) https://github.com/rbgirshick/py-faster-rcnn (python)
Multi-Person Pose estimation repository: https://github.com/ZheC/Realtime_Multi-Person_Pose_Estimation


Exercitation 21/03/2017

posted 20 Mar 2017, 11:24 by Simone Grazioso

For the tomorrow's exercitation, please bring your laptop with MATLAB and/or Python.
I will run an example of a CNN (using a modified version of caffe framework http://caffe.berkeleyvision.org ) that requires

  1. Matlab R2015a or higher
  2. OpenCV (2.4.xx or 3.0, https://goo.gl/mv0l0F )
  3. the latest (stable) nvidia drivers for your gpu
  4. possibly the cuda toolkit (https://developer.nvidia.com/cuda-toolkit). Is possible to run the examples in CPU mode, but is extremely slower.
  5. install ATLAS by sudo apt-get install libatlas-base-dev
  6. some general dependencies for caffe framework: sudo apt-get install libprotobuf-dev libleveldb-dev libsnappy-dev libopencv-dev libhdf5-serial-dev protobuf-compiler sudo apt-get install --no-install-recommends libboost-all-dev
  7. CUDNN (optional): (GPU-accelerated library of primitives for deep neural networks) https://developer.nvidia.com/cudnn
Note: CUDA v8 is required on Ubuntu 16.04, and the example that I will show you required about 2048mb of dedicated gpu memory.

Please download at least Cuda toolkit, because is a file of ~1.5gb and it's impossible to do it in class, due to the slow wifi

1-10 of 14