1. The following dataset is known as the DREAM data set. It is the most widely known dataset because of it shows how a child with ASD communicates differently depending on who they are talking to. It includes the behavioral data of 61 children diagnosed with ASD. Collected data comes from a large-scale study on autism therapy with the support of robots. The database includes over 3,000 sessions from more than 300 hours of therapy. Half of the children interacted with the social robot NAO, supervised by a therapist. The other half was considered as the control group and interacted directly with the therapist. Both groups followed the standard Applied Behavior Analysis (ABA). Each session was recorded with three RGB cameras and two RGBD cameras (Kinect) which were analyzed with an image processing technique to identify the child's behavior during the therapy. This public version of the database contains no recorded video material or other personal data, but instead includes anonymized data that describes the child's movements, head position and orientation, as well as eye movements, all specified in a common coordinate system. Furthermore, metadata is included in the form of the child's age, gender and autism diagnosis (ADOS). The full dataset is available through the following link: https://snd.gu.se/sv/catalogue/study/snd1156/1/1#dataset.

2.The following website contains several gene-related autism data sets from BioGPS, a reliable source for the datasets since it does a lot of research related to genes, so the datasets available are useful for genetics-autism research. It contains studies about specific genes such as FMR1: http://biogps.org/dataset/tag/autism/

​

 3. The following grid links to several research studies which contain several autism datasets.  https://catalog.data.gov/dataset/autism-prevalence-studies/resource/ff914e09-1409-444f-8663-44adb637a45c

​