2MLMD : Multi-Modal Leap Motion Dataset

A. Description : 

This Novel Public 2MLMD database has two modalities which are skeletal and depth data (see figure below). This database relates to 135 people.  The proposed dataset simplifies home control through 30 commands (24 dynamic and 6 static) executed using one or two hands through both modalities (skeletal and depth) delivered by a leap motion controller.

2MLMD is distinct due to its multi-modality and its extensive variety of gestures, involving various participants. Consequently, 2MLMD is categorized into three primary categories: 

• Static gestures performed with a single hand, 

• Dynamic gestures executed with one hand

• Gestures involving both hands.

Please cite the following papers if using this day dataset in your publications :  

To download this database, please send a request to :  anouar.benkhalifa@eniso.u-sousse.tn  

3MDAD : Multimodal Multiview and Multispectral Driver Action Dataset

A. Description : 

This Novel Public Dataset for Multimodal Multiview and Multispectral Driver Distraction Analysis is the result of collaboration between :

• Pr. Mohamed Ali MAHJOUB, Dr. Anouar BEN KHALIFA and Miss Imen JEGHAM from Laboratory of Advanced Technology and Intelligent Systems (LATIS)-National Engineering School of Sousse, Tunisia.

• Dr. Ihsen ALOUANI from the Polytechnic University Hauts-De-France.

The 3MDAD dataset is mainly composed of two real-world driving sets: the first one recorded at daytime and the second one at nighttime. Each set consists of two synchronised data modalities, each from frontal and side views. Two Microsoft Kinect cameras are employed. The first Kinect is mounted on the car handle on the top of the passenger window, and the second is placed on the dashboard in front of the driver.

• 3MDAD (Day) provides temporally synchronized RGB frames and depth frames. 50 participants (38 males and 12 females) aged between 19 and 41 were asked to drive the vehicle.

• 3MDAD (Night) provdes temporally synchronized IR frames and depth frames. 19 participants (11 males and 8 females) aged between 19 and 53 were asked to drive the vehicle. 

B. Action Classes : 

The actions that are commonly executed by individuals while driving are :

A1: Safe driving, A2: Doing hair and makeup, A3: Adjusting radio,  A4: GPS operating, A5: Writing message using right hand, A6: Writing message using left hand, A7: Talking phone using right hand, A8: Talking phone using left hand, A9: Having picture, A10: Talking to passenger, A11: Singing or dancing, A12: Fatigue and somnolence, A13: Drinking using right hand, A14: Drinking using left hand, A15: Reaching behind, A16: Smoking.

D. Citation :

Please cite the following papers if using this day dataset in your publications :  

Imen Jegham, Anouar Ben Khalifa, Ihsen Alouani, Mohamed Ali Mahjoub, A novel public dataset for multimodal multiview and multispectral driver distraction analysis: 3MDAD, Signal Processing: Image Communication, Volume 88, October 2020, 115966, DOI: https://doi.org/10.1016/j.image.2020.115960.   

Imen Jegham, Anouar Ben Khalifa, Ihsen ALOUANI, Mohamed Ali MAHJOUB, MDAD : A Multimodal and Multiview in-Vehicle Driver Action Dataset, In: Vento M., Percannella G. (eds) Computer    Analysis of Images and Patterns. CAIP 2019. Lecture Notes in Computer Science, vol 11679. Springer, Cham, pp. 518-529.  DOI: https://doi.org/10.1007/978-3-030-29888-3_42 

E. Contact :

If you have any questions or comments, please email : 

        - imen.jegham@isitc.u-sousse.tn

        - anouar.benkhalifa@eniso.u-sousse.tn

        - ihsen.alouani@uphf.fr

Infrastructure to Vehicle Multi-View Pedestrian Detection dataset (I2V-MVPD)

 This database is the result of collaboration between :

      - Pr. Mohamed Ali MAHJOUB and Dr. Anouar BEN KHALIFA from Laboratory of Advanced Technology and Intelligent Systems ((LATIS)-National Engineering School of Sousse, Tunisia.

      - Dr. Ihsen ALOUANI from the Polytechnic University Hauts-De-France.

A. Description : 

• I2V-MVPD dataset is the first real-world multi-view pedestrian detection database available to the scientific community combining both a static camera mounted on a road infrastructure and a mobile camera embedded in a vehicle’s dashboard. The database contains sixty-one sequences synchronically (Figure :  Illustration of the synchronization precision ) filmed by the mobile and the static cameras (and four static negative sequences) for a total of 4740 synchronized pairs of frames. The sequences depict a variety of simple and complex scenarios. 

• The proposed dataset is meant to be realistic and was designed for real-world applications. Hence, the sequences were filmed in an uncontrolled environment where other pedestrians and cars can enter and exit the filming locations freely.  The uncontrolled environment means also that the scene is cluttered with objects such as trees, light poles, signs and other fixtures.

• The first 20 sequences were fully annotated at 15 fps, and the other 41 sequences were annotated at a rate of 7.5 fps. Globally, 9480 frames were annotated with a total of 22127 bounding boxes (11164 bounding boxes in the static cameras and 10963 bounding boxes in the mobile cameras). 

B. Download details :

• Data from static camera fixed in a road infrastructure  :  Infrastructure (7.38 Go)

• Data from mobile camera embedded in a vehicle’s dashboard : Vehicle  (5.48 Go)

• Annotation data : infrastructure and  vehicle.

• Crops (Infrastructure  and Vehicle) 

To download this database, please send a request to :  anouar.benkhalifa@eniso.u-sousse.tn   

Please cite the following paper if using this dataset in your publications :  

Anouar Ben Khalifa, Ihsen Alouani, Mohamed Ali Mahjoub, Atika Rivenq, A novel multi-view pedestrian detection database for collaborative Intelligent Transportation Systems, Future Generation Computer Systems, Volume 113, December 2020, Pages 506-527, DOI: https://doi.org/10.1016/j.future.2020.07.025.

• Links to download:

  1. - Data samples for Class 0 (Background)  :  C_0 (0.5 Go)

  2. - Data samples for Class  1 (Pedestrian)  :  C_1 (4.4 Go)

  3. - Data samples for Class  2 (Cyclist)  :  C_2  (0.99 Go)

  4. - Data samples for Class  3 (Vehicle)  :  C_3 (1.82 Go)

  5. - Data samples for Class  4 (Tramway)  :  C_4 ( 0.64 Go)

  6. - Data samples for Class  5 (Several combinations of classes)  :  C_5 (7.58 Go)

To download this database, please send a request to :  anouar.benkhalifa@eniso.u-sousse.tn

Please cite the following paper if using this dataset in your publications :  

Amira Mimouna, Ihsen Alouani, Anouar Ben Khalifa, Yassin El Hillali, Abdelmalik Taleb-Ahmed, Atika Menhaj, Abdeldjalil Ouahabi, Najoua Essoukri Ben Amara, OLIMP: A Heterogeneous Multimodal Dataset for Advanced Environment Perception, Electronics, Volume 9, March 2020, 560. DOI: https://doi.org/10.3390/electronics9040560.  

LeapGestureDB

The dataset contains dynamic hand gestures performed by over 120 different people, each performing 11 different gestures repeated 5 times each. The Leap data consist in parameters provided by the Leap SDK (with the v2 version 2.3.1).

Our database is available as a set of 6600 text files click here to download the full dataset.

We provide the dataset used for paper. The dataset contains several different dynamic gestures acquired with the leap motion sensor. A part of this dataset has been exploited to test the recognition rate of a Multi-Class SVM gesture classifier.

To download this database, please send a request to :  anouar.benkhalifa@eniso.u-sousse.tn

Please cite the following papers if using this day dataset in your publications :  

 Safa Ameur, Anouar Ben Khalifa, Mohamed Salim Bouhlel, LeapGestureDB: A Public Leap Motion Database Applied for Dynamic Hand Gesture Recognition in Surgical Procedures, In: Balas V., Jain L., Balas M., Shahbazova S. (eds) Soft Computing Applications. SOFA 2018. Advances in Intelligent Systems and Computing, vol 1222. pp. 125-138, Springer, Cham. DOI: https://doi.org/10.1007/978-3-030-52190-5_9  

Safa Ameur, Anouar Ben Khalifa, Mohamed Salim Bouhlel, A comprehensive leap motion database for hand gesture recognition, 7th International Conference on Sciences of Electronics, Technologies of Information and Telecommunications, pp. 514 - 519, 2016. DOI: https://doi.org/10.1109/SETIT.2016.7939924 

Electrocardiogram: LATIS ECG DATABASE:

Over  the  last  few  years,  the  Electrocardiogram (ECG)  was  introduced  as  a  powerful  biometric  modality  for human  authentication.  Indeed, ECG has some characteristics specific to each individual. The  potential  benefit  of  transmitting ECG  in  the  field  of  biometrics  is  that  it  is difficult  to  be spoofed or falsified. Unlike other traits that are unique, but not secure. For example a face can be falsified by a mask, the iris by a lens, and the voice by a record.  Furthermore,  the ECG  signal  provides  evidence  for  aliveness,  is  a permanent modality  and it’s unique for  each  person.

The LATIS ECG DATASET contains 6 Derivation (D1, D2, D3, aVR, aVL, aVF) performed by 100 volunteers (61 female and 39 male). The different people who participated in the data collection have different ages and cultural levels. The database is used for biometric identification of persons.

To download this database, please send a request to :  anouar.benkhalifa@eniso.u-sousse.tn

Click here (4 MB)to download the LATIS ECG DATASET (from researchgate data). Any publication using this database must cite the following paper :

Takoua Hamdi, Anis Ben Slimane, Anouar Ben Khalifa, A novel feature extraction method in ECG biometrics, Conference International Image Processing, Applications and Systems, pp. 1-5, 2014.

LATIS OFF LINE HANDWRITING and SIGNATURE DATABASE

The biometric authentication rests on the exploitation of characteristics or measurements related to the physiology and/or the behaviour of one person (voice, face, signature, writing, fingerprint, gait…) for the checking or the authentication of his/her identity. Each one of these characteristics is called "biometric modality". The use of the off line handwritten signature and handwriting for establishing an individual’s identity was the object of numerous studies. These two modalities have the advantage of being easy acquired and socially well accepted. The context of their use is diversified: banking environment, financial, commercial, legal…Nevertheless, the exploitation of these two methods is not commonplace, in particular because of the problems of variability within and between persons. Indeed, as far as the shape of the signature and the writing of the same person differ due to different factors such as the position of one’s hand, his/her moral and emotional state, his/her physical form and other types of constraints such as the temporal constraints…, from which very significant within-writer variations often encountered (figure 1.a). In addition, these two modalities change considerably from one person to another (figure 1.b). These between-writer variations are related to various cultural and individual aspects as age, sex, school level, technical conditions…

Figure 1.  Illustration of within-writer and between-writer by the superposition of the same word and the same signature:

(a) written 5 times by the same person and (b) written by 5 different persons.

Description:

LATIS OFF LINE HANDWRITING and SIGNATURE DATABASE is relating to 100 people. Each person provided 60 samples of his/her signature, name and date and place of birth. So we had 12000 samples of signatures and handwritings.

The different people who participated in the data collection have different ages and cultural levels. Figure 2 shows some sample extracts of our database.

 To download this database, please send a request to :  anouar.benkhalifa@eniso.u-sousse.tn

Figure 2. Examples of image extracts of our database

 Click here (50.78 MB) to download the LATIS OFF LINE HANDWRITING and SIGNATURE DATASET (from researchgate data). Any publication using this database must cite the following paper :

Anouar Ben Khalifa, N. Essoukri BenAmara. Bimodal biometric verification with different fusion levels, IEEE International Multi-Conference on Systems, Signals & Devices, Djerba, Tunisia, pp.1-6, 2009.