The poultry industry has been driven primarily by broiler chicken production and has grown into the world’s largest animal protein sector. Automated detection of chicken carcasses on processing lines is vital for quality control, food safety, and operational efficiency in slaughterhouses and poultry processing plants. However, developing robust deep learning models for tasks like instance segmentation in these fast-paced industrial environments is often hampered by the need for laborious acquisition and annotation of large-scale real-world image datasets. To this end, we present the first pipeline generating photorealistic, automatically labeled synthetic images of chicken carcasses under varying poses. We also introduce a new benchmark dataset containing 300 annotated real-world images, curated specifically for poultry segmentation research. Using these datasets, this study investigates the efficacy of synthetic data and automatic data annotation to enhance the instance segmentation of chicken carcasses, particularly when real annotated data from the processing line is scarce. A real dataset (300 images of chicken carcasses) with varying proportions of synthetic images were evaluated in prominent instance segmentation models: You Only Look Once version 11 segmentation (YOLOv11-seg), Mask Region-based Convolutional Neural Network (Mask R-CNN) (with R50 and R101 backbones), Masked-attention Mask Transformer (Mask2Former), and a finetuned segment anything model (SAM) [Vision Transformer (ViT)-B]. Results demonstrate that synthetic data significantly boosted segmentation performance for YOLOv11-seg, Mask R-CNN, and Mask2Former. YOLOv11-seg consistently achieved the highest segmentation accuracy. Notably, models with greater capacity (R101) and transformer-based architectures (Mask2Former) derived greater benefits, particularly with larger volumes of synthetic data. Conversely, while SAM’s segmentation degraded, its bounding box detection performance paradoxically improved with synthetic data, achieving the highest detection scores among all models. Furthermore, model-specific optimal ratios of synthetic-to-real data were observed. This research underscores the value of synthetic data augmentation as a viable strategy to mitigate data scarcity and reduce manual annotation efforts, and advance the development of robust AI-driven automated detection systems for chicken carcasses in the poultry processing industry.