Facial Emotion Recognition Images for Workplace Analysis, 35,685 Grayscale Faces

A dataset of 35,685 grayscale facial images across seven emotion categories, created by Makhan Kumbhkar and last updated in May 2026. It is derived from the FER-2013 dataset and is intended for research on the relationship between employee emotions and work culture. The accompanying model uses MobileNet with Transfer Learning and OpenCV Haar Cascade for face detection.

Use Cases

• Training real-time facial emotion detection models based on the described MobileNet architecture.
• Analyzing the relationship between employee emotions and organizational work culture as mentioned in the description.
• Developing systems for mental health monitoring or educational engagement using facial expression data.
• Creating tools to inform HR policy by leveraging a Happiness Index, as suggested in the description.

Strengths

• Contains 35,685 pre-labeled grayscale facial images.
• Covers seven distinct emotion categories: happiness, sadness, anger, disgust, fear, surprise, and neutral.
• Includes a pre-trained model using MobileNet with Transfer Learning and OpenCV Haar Cascade for face detection.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• The geographic and demographic origin of the facial images is not specified, which may indicate source bias.

Provenance

Source

Derived from the FER-2013 dataset.

Collection Method

Images were likely gathered and labeled as part of the original FER-2013 collection process.

Freshness

Last updated 2026-05-22 08:43:25; freshness should be verified.