This paper has described the adaptability of semantic segmentation using deep convolutional neural networks for detecting forest strip roads. We used fully convolutional networks, which are one of the most commonly used methods in semantic segmentation. We established our datasets by capturing videos of strip roads. We performed a 3-class classification of strip roads, their backgrounds, and the buffer between the roads and the backgrounds and achieved good overall accuracy (96.7–97.2%). Our findings showed that our approach can detect strip roads to the same level as in the previous research. Our precision for the 3classes of strip road, buffer, and background was 96.7–97.5%, 67.4–71.6%, and 98.5–98.6%, respectively. There was some difficulty in detection of the edge part of strip roads as a buffer class using the model in this study. However, our trained model showed good precision in detecting the strip road class. These findings suggest that our models might predict strip roads for local path planning when strip roads are sufficiently wide for machines to pass.

In this study, we define sustainable forest utilization based on forest information and assess the productivity of harvesting process of forestry machinery for sustainable utilization of forests in the Oku Aizu region of Fukushima Prefecture. First, we look at the case of the Nishikata area in Mishima town and analyze its labor productivity. Then, we consider the adaptation of the harvesting system used in the Nishikata area throughout the Oku Aizu region and evaluate the number of personnel required for sustainable forest utilization. The productivity assessment of the harvesting system in the Nishikata area provided an estimate of 3.3–8.3 m ³ /person-day in the steeply sloping forests. Productivity in the gently sloping forests was estimated to be 38–53 m ³ /person-day. Sustainable forest utilization in the Oku Aizu region requires harvesting and reforesting 160–180 ha/year. The number of persons needed for harvesting in Oku Aizu was 34–110 in the first 50 years and 34–76 thereafter. Using the suggested process, productivity in the gently sloping forests in the Nishikata area was similar to that in Austria and other countries with developed forestry. Meanwhile, the productivity of the steeply sloping forests was either equal to or higher than that in Japan at present. Measures to improve productivity in countries with industrial forestry necessitate expanding the forest road network and revising working hours.

This study used an“onion seedling transplanter”to reduce the labor associated with transplanting Japanese cedar seedlings. This method allowed approximately 1800 seedlings to be transplanted per hour per worker compared with approximately 150–180 seedlings per hour per worker using the hand planting method. This indicated that use of the machinery enhanced the work efficiency by > 10 times. The appropriate cell size of seedlings for the machine was approximately 5–15 cm. We found that the largest proportion of seedlings met this standard when they were sown in the middle of May. The seedlings were then sorted in the middle of August, and seedlings that were > 10 cm in height were sprayed with the growth suppressor Ethrel 10, whereas those with < 5 cm in height were given additional fertilizer treatment to promote their growth. This treatment resulted in approximately 90% of the seedlings meeting the standard of the machinery. Second-year Japanese cedar seedlings are generally transplanted in spring. However, to distribute the nursery-related work throughout the year, we transplanted the seedlings in the fall of the previous year, which resulted in a survival (rooting) rate of approximately 95%. Therefore, shifting the transplanting to fall allowed the annual labor to be more evenly distributed.

This study aimed to investigate the effect of travel distance out of cutting areas on the productivity of forestry machinery. We measured the travel distance and the shortest distance in straight lines between the cutting area and the forest road using a forest road map from the national forests in Hokkaido. The results showed that the forwarder does not travel on the forest roads when access roads to the cutting area are 1500 m or longer than the shortest distance determined from the forestry map. In addition, running distances on access roads increase until reaching 1000 m. The aim of the study was to achieve travel distances of less than 500 m in the national forests of Hokkaido. Establishing a cutting area within 50 m of the shortest distance access the road achieves this aim while decreasing the travel distance out of the cutting area. The result is a decrease of up to 7% in the forwarder productivity.